Article Text

Download PDFPDF

What’s on your keyboard? A systematic review of the contamination of peripheral computer devices in healthcare settings
  1. Nicole Ide1,
  2. Bianca K Frogner1,
  3. Cynthia M LeRouge2,
  4. Patrick Vigil3,
  5. Matthew Thompson1
  1. 1 Department of Family Medicine, University of Washington, Seattle, Washington, USA
  2. 2 Department of Information Systems & Business Analytics, Florida International University, Miami, Florida, USA
  3. 3 Family Medicine, Pacific Northwest University, Yakima, Washington, USA
  1. Correspondence to Dr Matthew Thompson; mjt{at}uw.edu

Abstract

Objective To determine the extent and type of microbial contamination of computer peripheral devices used in healthcare settings, evaluate the effectiveness of interventions to reduce contamination of these devices and establish the risk of patient and healthcare worker infection from contaminated devices.

Design Systematic review

Methods We searched four online databases: MEDLINE, CINAHL, Embase and Scopus for articles reporting primary data collection on contamination of computer-related equipment (including keyboards, mice, laptops and tablets) and/or studies demonstrating the effectiveness of a disinfection technique. Pooling of contamination rates was conducted where possible, and narrative synthesis was used to describe the rates of device contamination, types of bacterial and viral contamination, effectiveness of interventions and any associations between device contamination and human infections.

Results Of the 4432 records identified, a total of 75 studies involving 2804 computer devices were included. Of these, 50 studies reported contamination of computer-related hardware, and 25 also measured the effects of a decontamination intervention. The overall proportion of contamination ranged from 24% to 100%. The most common microbial contaminants were skin commensals, but also included potential pathogens including methicillin-resistantStaphylococcus aureus, Clostridiumdifficile, vancomycin-resistantenterococci and Escherichia coli. Interventions demonstrating effective decontamination included wipes/pads using isopropyl alcohol, quaternary ammonium, chlorhexidine or dipotassium peroxodisulfate, ultraviolet light emitting devices, enhanced cleaning protocols and chlorine/bleach products. However, results were inconsistent, and there was insufficient data to demonstrate comparative effectiveness. We found little evidence on the link between device contamination and patient/healthcare worker colonisation or infection.

Conclusions Computer keyboards and peripheral devices are frequently contaminated; however, our findings do not allow us to draw firm conclusions about their relative impact on the transmission of pathogens or nosocomial infection. Additional studies measuring the incidence of healthcare-acquired infections from computer hardware, the relative risk they pose to healthcare and evidence for effective and practical cleaning methods are needed.

  • infection control
  • contamination
  • healthcare-acquired infection
  • nosocomial infection
  • keyboard

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

View Full Text

Statistics from Altmetric.com

Strengths and limitations of this study

  • This is the first systematic review on the level of contamination of computer peripheral devices used in clinical care as well as the effectiveness of interventions used to decontaminate these surfaces.

  • We searched four major online databases during the literature search and hand searched references of included studies and relevant review articles.

  • Reporting of this review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.

  • The ability to perform meta-analysis was limited by the heterogeneity among the included studies.

Introduction 

The annual number of healthcare-acquired infections (HAIs) in the US acute care hospitals is estimated at approximately 722 000, or 4% of inpatients.1 HAIs lead to longer admissions, more frequent readmissions and poorer patient outcomes including increased mortality.2 3 The US Centers for Disease Control and Prevention (CDC) estimates that preventing HAIs in the USA would result in annual direct savings of between US$5.7 and US$31.5 billion.4 Studies to date have largely focused on hospital settings; thus, the frequency of consequences of HAIs in outpatient settings is poorly described.

Between 20% and 40% of HAIs result from cross-infection via hands of personnel, and another 20% from other environmental contamination.5 Contamination of environmental surfaces in healthcare settings is a well-known source of nosocomial infection, and several pathogens have been identified on surfaces in hospital environments, including methicillin-resistant Staphylococcus aureus (MRSA), Clostridium difficile (C. diff), Acinetobacter baumannii, vancomycin-resistant enterococci (VRE), Pseudomonas aeruginosa, Norovirus and Gram-negative bacteria.6–9 Nosocomial pathogens often originate from infected patients who come into contact with the surfaces surrounding them, particularly ‘high-touch surfaces’, and are then transferred to other healthcare workers’ or patients’ hands.

Several studies looking at healthcare workers’ personal devices (mobile phones or personal digital assistants (PDAs)), clothing (neckties, white coats, etc) and a variety of other objects (stethoscopes, blood pressure cuffs, telephones, faucets, bedrails, etc) have found significant rates of environmental contamination.6 10 11 However, the importance of contamination related specifically to computer keyboards, mice and other computer peripherals is less well established despite their ubiquitous use in hospital and ambulatory healthcare settings.

We, therefore, conducted a systematic review to determine the extent to which computer keyboards, mice and other computer peripheral devices have been identified as being a source of contamination in clinical settings. We examine the type and prevalence of microbial contamination, and the settings in which these contaminated devices have been addressed. We also determined the effectiveness of interventions that aim to reduce contamination of these devices, and any evidence linking clinical consequences of HAI related to computer keyboards/peripherals among patients and healthcare workers.

Methods

We report this systematic review in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, an evidence-based minimum set of items recommended for reporting of systematic reviews.12 A PRISMA checklist can be found in online supplementary file 1.

Supplementary material 1

Search strategy

A total of four databases were included in our search: MEDLINE, CINAHL, Embase, and Scopus. We developed two major categories of search terms that were used in various combinations to search the databases. First, terminology related to peripheral and external computer hardware devices, such as mice and keyboards. Second, terminology related to infection, contamination or disinfection (online supplementary file 2). We conducted automated searches in databases from 01 January 1990 to 14 July 2017. We limited the search to this time frame due to the low rates of computer use in clinical settings prior to 1990. Additionally, we manually searched the references of included studies and relevant review articles to identify further eligible studies, and where possible, we contacted authors to obtain full texts of abstracts if not available online.

Supplementary material 2

Eligibility criteria and study selection

We included studies that met the following criteria: (A) conducted in any type of healthcare setting in a high-income or upper-middle-income country,13 (B) investigated keyboards, mice, mouse pads, computer touch screens, laptops and iPads/tablet computers, (C) reported primary data collected through experimental, quasi-experimental or observational study designs, (D) reported contamination rates of computer-related equipment and/or demonstrated the effectiveness of disinfection technique(s), (E) reported any association between contamination of computer-related equipment and infection or colonisation of patients/healthcare workers and (F) written in English language.

We excluded studies that were not conducted in a healthcare setting or were conducted in low-income or lower-middle-income countries (where pathogenic microbes are potentially different to those found in high-income or upper-middle-income countries), tested computer-related equipment with in vitro experiments, reported solely data on environmental surfaces other than computer-related hardware, or assessed healthcare worker knowledge or compliance with disinfection or hand-washing protocols. We excluded all studies that only provided an abstract.

After searching the four databases, we uploaded articles to EndNote X8 and removed any duplicates. One reviewer (NI) screened titles and abstracts to remove clearly irrelevant studies. Two reviewers (NI and MT) independently screened the full text of all remaining articles to determine final eligibility, and resolved any discrepancies through discussion and consensus.

Data extraction and quality assessment

Using a standardised form in Microsoft Excel, a single reviewer (NI) extracted the following data from each included article: country and clinical setting, study design, sampling frame and size, microbiological sampling method, microbiological identification method, outcome measure(s), intervention definition (if any), comparison (if any), ongoing decontamination methods (if any) and results (baseline contamination rates, baseline pathogens detected and post-intervention contamination rate). Extracted data were checked for accuracy by a second author (MT), and disagreements were resolved prior to analysis.

Two authors (NI and MT) independently assessed the methodological quality and risk of bias using checklists we developed based on The National Heart, Lung, and Blood Institute’s study quality assessment tool14 as well as criteria developed in a relevant systematic review by Livshiz-Riven et al, which assessed the relationship between contamination and non-invasive portable clinical environmental surfaces.15 To assess the risk of bias for each outcome, we developed two separate checklists: one for studies reporting only baseline contamination and another for studies that included an intervention. We looked at the quality of individual studies and assessed the risk of bias on the basis of study design, objectives, sampling strategy, microbial detection methods, outcome measurement and reporting, and confounding variables. For studies of decontamination interventions, we also assessed intervention characteristics and comparisons or controls. Each assessment item was scored as ‘Yes’, ‘No’ or ‘Unclear’. The overall risk of bias of the body of evidence was considered in the interpretation of findings of the review.

Summary measures

For studies reporting contamination of peripheral computer-related hardware devices, we present findings as the proportion of devices contaminated, using definitions of contamination as reported in individual studies. For studies reporting the effectiveness of a decontamination intervention, we present findings as a change (or percentage change) in contamination rates following the intervention, as reported by the respective authors. We explored whether there were differences in contamination rate between clinical settings, countries or types of devices. We intended to use meta-analysis to pool results, but due to heterogeneity in study design, interventions and outcomes reported, this was not possible. A simple pooled mean of baseline contamination of the studies, which included an overall baseline rate of device contamination, was calculated.

Patient and public involvement

Neither patients nor the public was involved in the development of the research question or study design for this systematic review. Results will be made available to the public by publishing this study in a peer-reviewed, open access journal.

Results

Study selection

Our search identified 4416 records, with an additional 24 identified through a manual search. After removing duplicates, we screened the remaining 3920 articles based on our inclusion criteria. Of these, 174 were selected for full-text review, of which 99 did not meet our criteria and were excluded, leaving a total of 75 studies in the final analysis (figure 1).16–90

Figure 1

Flow diagram of study selection.

Study characteristics

Of the 75 included studies (online supplementary file 3), only one was published prior to the year 2000, with another 27 studies published between 2000 and 2009, and 47 studies published 2010 onwards. Most were conducted either in the USA or Canada (26) or Europe/Central Asia (28), followed by Southeast/East Asia or the Pacific (12), Middle East (4), South America (4) and South Africa (1).

Supplementary material 3

The vast majority (63) of studies were conducted only in hospitals, including intensive care units (ICUs) (12 conducted solely in ICU and an additional 17 studies included ICU as one of their settings), emergency department (11) and operating rooms (8). Twelve studies were conducted in a variety of other clinical settings, including dental clinics or a dental hospital, radiology settings, an outpatient ophthalmology clinic, a pharmacy practice or a mixed setting.

Overall, the included studies provided data on a total of 2804 devices, including 1482 keyboards, 665 computer stations and 398 mice or mouse pads. Nineteen studies did not explicitly state the number of devices tested or only reported the total number of samples taken. Keyboards were the most commonly studied peripheral computer device, with 42 studies testing keyboards alone and another 22 testing a combination of keyboards plus mice. Fewer tested tablets (5) or mice alone (2). The numbers of devices sampled ranged from a single keyboard up to 282 computer stations (keyboards plus mice).

The majority of studies (50) reported primarily on device contamination rates (mostly using cross-sectional samples).17–23 26 29 32–36 38 41–46 49 50 52–56 60 62 64–66 68–76 81–86 90 Another 25 studies used interventional designs16 24 25 27 28 30 31 37 39 40 47 48 57–59 61 63 67 77–80 87–89; most reported contamination rates before and after a disinfection or cleaning process (and therefore also contributed data on baseline contamination rates). One study only reported contamination postintervention,61 and another two reported only on an association between device contamination and patient colonisation rates.63 88 Of the 25 studies reporting interventions, most used pre–post designs (17), with a smaller number (8) using controlled trials, postintervention study, cross-over or prospective comparative analysis. A variety of methods were used to measure effectiveness, including change in rate of overall contamination (11), change in rate of specific pathogens (5), change in colony forming unit (CFU) values (3), reduction in both rates and CFU values (2), rate of keyboards with contamination over 500 CFU (1), number of acquired colonisations preintervention and postintervention (1), patient acquisition of MRSA (1) and contamination rate for postintervention phase only (1).

Prevalence of baseline contamination

A total of 71 studies provided data on levels of device contamination. Of these, 26 presented an overall proportion of microbial contamination (table 1), with contamination rates ranging from 24% to 100%. Of these 26 studies, 21 reported the proportion of devices contaminated, while five reported the proportion of collected swabs that were contaminated. Of the 21 studies reporting device contamination, the pooled mean contamination rate was 96.7% (range 80% to 100%).

Table 1

Studies reporting the proportion of computer devices contaminated

A further 12 studies reported overall contamination only as CFU (online supplementary file 4), and another 10 reported contamination using a variety of other methods, such as proportion of devices with multiple bacterial species identified, mean bacterial counts, aerobic colony counts or ATP values/failures (online supplementary file 5). A further 23 studies reported baseline contamination of only a single or few specific pathogens: 20 as a proportion (%) of each pathogen, one presented total bacterial counts (mean±SD) and two reported the existence of specific pathogens without quantifying them (online supplementary file 6).

Supplementary material 4

Supplementary material 5

Supplementary material 6

The range of overall contamination was wide: while most studies found a contamination rate of 80%–100%, Bures et al reported a rate of 24% in a study of keyboards in ICU patient rooms and nurse/doctor stations,20 while Smith et al reported a rate of 43% on notebook computers from medical, surgical and family practice programmes.78 However, we were unable to determine differences in contamination rates between clinical settings, countries or types of devices due to insufficient data.

Type of microbial contamination

The specific pathogens isolated from keyboards or other computer devices was reported in 63 studies. Of these, 49 reported the proportion of devices contaminated with specific types of bacteria (online supplementary file 7). The most frequent microbial contaminants were skin commensal bacteria, but contamination with a variety of potentially pathogenic bacteria was also reported. The most frequent potential pathogens identified included Staphylococcus aureus (S. aureus) and MRSA, but this depends on whether studies set out to detect all microbe or pathogens, or only specific organisms. Of the studies reporting contamination with S. aureus, the mean contamination rate was 28% (range 1%–94%). Mean rates of contamination with MRSA was 14% (range 0%–100%), VRE 3.7% (range 0%–12%) and C. diff 8.0% (range 0%–28%).

Supplementary material 7

Effectiveness of decontamination interventions

Twenty-five studies evaluated the effectiveness of disinfection or cleaning interventions on the level of device contamination. Of these, 14 reported statistically significant reductions in contamination following the intervention (table 2). These included seven studies using wipes/pads with isopropyl alcohol, quaternary ammonium, chlorhexidine or dipotassium peroxodisulfate16 24 31 37 47 67 89; three studies using ultraviolet (UV) light39 57 77; two studies using putty cleaning compound58 59; one study with an enhanced cleaning protocol (including glove use)63 and one study using a keyboard with a cleaning alarm.87

Table 2

Studies reporting interventions that led to a significant reduction in contamination of computer peripheral devices

A further eight studies reported reductions in contamination from interventions (online supplementary file 8), but reductions were not statistically significant,78 not tested using statistical tests,28 48 79 80 or did not apply the statistical tests specific to data from the computer devices.27 30 40 Effectiveness of interventions in an additional two studies was unclear due to poor reporting of baseline and/or postintervention contamination rates (online supplementary file 8).25 61

Supplementary material 8

Association between device contamination and clinical infection

Only five included studies examined the association between device contamination and infection or colonisation of patients/healthcare workers (online supplementary file 9). Of these, three reported an association showing that the decontamination intervention was associated with reductions in the rate of MRSA infections,27 VRE40 and Acinetobacter colonisations.63 However, the link between association and causation in these studies was unclear and open to bias. One study showed that even though 12.5% of positive blood cultures matched the organisms growing from surveillance sites, this correlation was not significant,70 and one showed no effect of a cleaning intervention on patient acquisition of MRSA.88

Supplementary material 9

Quality assessment

For studies that reported contamination rates, sampling methods were often convenience-based, and only six used a power calculation to guide sample size. In 19 studies, the number of included devices was not explicitly stated, and denominators were reported inconsistently. In 44 out of 75 studies, selection criteria for the devices were not given and not clearly described or implemented consistently. In 29 of the 50 studies that only measured prevalence, samples were obtained at a single time point. Only four of the studies that reported effectiveness of decontamination interventions were controlled trials, with most using cross-sectional or pre–post designs. Reporting of the effectiveness of interventions using statistical testing was poor or inconsistent. Few studies were designed in such a way that patient outcomes could be measured, that is, the direct impact of contamination on HAI. Reporting of results was frequently poor, with only 26 studies reporting the overall number and percentage of computer-related devices with bacterial contamination. Of the 50 studies reporting only baseline contamination, only 10 studies provided a CI or mean/median CFU, ATP or relative light unit value of keyboards or computer peripherals sampled. Full risk of bias tables can be found in online supplementary file 10

Discussion

To the best of our knowledge, this is the first systematic review to report on the level of contamination of computer peripheral devices used in healthcare settings, as well as the effectiveness of interventions used to decontaminate these items. This review fills an important gap and provides substantial evidence from 75 studies and a total of 2804 devices, that is, computer peripheral devices, particularly keyboards, are potential reservoirs of infective pathogens. The overall proportion of contamination ranged from 24% to 100%. Collectively, studies found a 96.7% contamination rate of keyboards sampled. Keyboards and other computer peripherals were most commonly contaminated with skin commensal bacteria, but also with a variety of other potential pathogenic bacteria including MRSA, C. diff, VRE and E. coli. Multiple interventions have been tested in attempts to decontaminate computer devices and keyboards in clinical settings, and several appear effective at reducing the overall level of contamination. Fourteen of the 25 interventional studies reported statistically significant reductions in contamination following the intervention. Effective interventions include: wipes/pads using isopropyl alcohol, quaternary ammonium, chlorhexidine or dipotassium peroxodisulfate, UV light emitting devices, putty cleaning compounds, enhanced cleaning protocols and a keyboard with a cleaning alarm. However, results were inconsistent and there was insufficient data to provide robust recommendations on which method(s) are most effective to adopt routinely. Finally, there was insufficient data to demonstrate clear evidence of an association between contamination and human infection.

Current data are mostly limited to hospital settings. Almost all (63) of the included studies were conducted solely in hospitals, with a particular focus on ICUs. Only a small number of studies were conducted solely in ambulatory or outpatient settings.

Comparison with the existing literature

Our findings are consistent with a variety of literature on the potential contribution of contaminated hospital surfaces to human infection.91 Not only can environmental surfaces harbour dangerous pathogens, but evidence shows that pathogens, such as MRSA, can be transferred to healthcare workers’ gloves or hands from contaminated surfaces.92–94 While some pathogens only survive a few days on inanimate surfaces, others, such as VRE, MRSA, Acinetobacter spp. and C. diff can survive for months if not properly cleaned or disinfected.95 96 Furthermore, some pathogens, such as VRE or C. diff, are more resistant to common disinfection methods than others. The link between environmental contamination and human infection has been difficult to establish firmly; however, various modelling studies, observational epidemiological studies, interventional studies, as well as outbreak reports suggest that this link exists.7 97 98

The optimal strategies for environmental disinfection in healthcare settings is unclear. Substantial evidence suggests that relying only on hand hygiene compliance among health workers is not an effective strategy. Two systematic reviews showed median rates of compliance with hand hygiene guidelines in hospital settings of 40%–57%.99 100 Keyboards and computer devices pose additional challenges, including the difficulty of decontaminating their irregular surfaces and the potential for damage from cleaning products.101 While multiple methods to decontaminate environmental surfaces generally have been developed, but their effectiveness is unclear.96 98 102 103 Indeed, the CDC’s Guidelines for Environmental Infection Control in Health-Care Facilities (updated in 2011) concluded that ‘more research is required to clarify the effectiveness and reliability of fogging, UV irradiation, and ozone mists to reduce norovirus environmental contamination,’ giving it a ‘No recommendation/unresolved issue’ rating.104 Results from our review suggest that little progress has been made in providing robust evidence for decontamination methods.

Limitations of the review

As with any systematic review, our findings are limited by the quantity and quality of the included studies. Heterogeneity across a number of areas limited our ability to conduct a meta-analysis and/or draw inferences from our findings. This included heterogeneity in the swabbing and microbiological identification methods, study settings, study timeframes, sample sizes and types of included devices. Outcome measures also varied; for example, some studies did not report a baseline contamination rate, and others did not specify the prevalence of specific pathogens identified. Fewer than half of the studies reported selection criteria that were prespecified, clearly described and implemented consistently. Only one study specifically sought to identify viruses (norovirus).61 Many potential pathogens were not specifically assessed in the included studies, and the data may represent an underestimate of contamination rates. Finally, nearly all included articles were conducted in hospital environments, and we have limited data on ambulatory or primary care settings.

Implications for researchers, clinicians and policymakers

Our findings indicate that the majority of keyboards and computer peripherals used in healthcare settings are contaminated with a range of microbes, including potential pathogens. However, determining the impact of this contamination on patients or healthcare workers was limited. Although we searched for studies reporting associations between contamination of computer-related equipment and infection or colonisation of patients/healthcare workers, very few studies (5) were identified and the results of these were unclear and open to bias. Thus, our findings do not allow us to draw firm conclusions about the relative impact of these ‘reservoirs’ of contamination as sources of transmission between patients and healthcare staff, nor their impact on HAI or nosocomial infections. However, given that computers are ubiquitous in modern healthcare, it is possible that keyboards and peripherals may act as important, yet largely unrecognised sources of contamination and/or infection. Although evidence directly linking contaminated computer equipment and HAIs is scarce, evidence does demonstrate the effectiveness (although sometimes limited) of decontaminating potential fomites other than computer equipment as well as health workers’ hands on reducing HAIs.7 97 98 105–107 Given this evidence, there is an urgent need to identify whether the same benefits apply to decontaminating computer equipment.

Our review highlights priorities for further research in this area. First, there seems to be little need to further demonstrate the prevalence of contamination on computer-related devices. In contrast, however, the relative impact of computer device contamination on colonisation and infection of patients/healthcare workers is unclear currently; thus, future research should focus on clinically significant organisms and their potential for transmission to patients or health workers. Additionally, more robust study designs are needed for evaluating decontamination interventions, particularly ones that could be used in routine practice.

In conclusion, computer keyboards and other peripheral computer devices in hospital settings are frequently contaminated, often with potentially pathogenic microbes. It is unclear from current research how often these lead to HAI, and what measures clinicians and their staff should take (and how often) to ensure that their computers are sufficiently clean and do not pose risks for themselves or their patients.

Supplementary Table 1

Supplementary Table 2

Acknowledgments

Many thanks to Sarah Safranik for assisting with the search strategy for the literature review.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
  106. 106.
  107. 107.
View Abstract

Footnotes

  • Contributors NI and MT designed the study methodology and conducted the literature search. NI extracted data from selected studies and MT checked extracted data for accuracy. NI and MT performed data analysis and developed the original draft of the article, and contributed towards further drafts. Data interpretation and critical revision of the manuscript was done by BKF, CML and PV. All authors reviewed and approved the manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The complete data extraction form, quality assessment tables, and full search strategy can be made available upon request to the study authors.

  • Patient consent for publication Not required.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.