Types of studies

We planned to include RCTs and quasi‐RCTs (if sufficient evidence was provided to demonstrate that the treatment and control groups are similar at baseline).

In reference to cross‐over trials, as we were currently unaware of the long‐term effects of these interventions, we were unable to determine whether the effects of the first intervention will interfere with those of the second. In order to avoid introducing this bias into the analysis, we planned to include only first‐arm data from cross‐over trials, when available.

Types of participants

People with known SCD (SS, SC, Sβ⁺ and Sβ⁰, proven by electrophoresis and sickle solubility test, with family studies or DNA tests as appropriate) of all ages and both sexes, in any setting.

Types of interventions

We included all standard modes of TENS, biphasic or monophasic electrical current delivered in pulses in high intensity or high frequency, low intensity or low frequency, or other standard modes that produce perceptible sensation at the area of application. We excluded other modes of electrotherapy, e.g. TENS delivered in barely perceptible intensity.

Eligible comparisons are:

• TENS with conventional treatment (e.g. analgesics) versus conventional treatment alone;

• TENS versus placebo (sham TENS);

• TENS versus other non‐pharmacological modalities for managing pain.

We did not plan to compare different intensities and frequencies of TENS.

Placebo (sham TENS) devices may look exactly the same as active TENS devices, but are deactivated and produce no current or may produce a brief period of stimulation at the beginning which fades out later. Due to the lack of perceptible stimulation in sham TENS, the blinding of participants to the mode of treatment is almost impossible (Sluka 2013) and this represents a risk of bias to all sham‐controlled trials of TENS (Gibson 2015).

Types of outcome measures

Electronic searches

The Cochrane Cystic Fibrosis and Genetic Disorders Group's Information Specialist conducted a search of the Group's Cystic Fibrosis Trials Register for relevant trials using the following terms: (sickle cell OR (haemoglobinopathies AND general)) AND TENS).

The Haemoglobinopathies Trials Register is compiled from electronic searches of the Cochrane Central Register of Controlled Trials (CENTRAL) (updated each new issue of the Cochrane Library) and weekly searches of MEDLINE. Unpublished work is identified by searching the abstract books of five major conferences: the European Haematology Association conference; the American Society of Hematology conference; the British Society for Haematology Annual Scientific Meeting; the Caribbean Health Research Council Meetings; and the National Sickle Cell Disease Program Annual Meeting. For full details of all searching activities for the register, please see the relevant section of the Cochrane Cystic Fibrosis and Genetic Disorders Group's website.

Date of the most recent search of the Cystic Fibrosis and Genetic Disorders Group's Haemoglobinopathies Trials Register: 26 February 2020.

We searched the following databases (Appendix 2):

• PubMed (www.ncbi.nlm.nih.gov/pubmed, 1946 to 04 April 2017);

• Embase Ovid (1974 to 16 October 2017).

In addition, we also searched the following trial registries and other resources (Appendix 2):

• US National Institutes of Health Ongoing Trials Register, ClinicalTrials.gov (clinicaltrials.gov/; searched on 04 September 2017);

• International Standard Randomised Controlled Trials Number (www.isrctn.com/; searched on 04 September 2017);

• World Health Organisation International Clinical Trials Registry Platform (apps.who.int/trialsearch/; searched on 04 September 2017);

• Google Scholar (scholar.google.co.uk/; searched on 04 September 2017).

Regarding our use of Google Scholar, we performed a Google Scholar Advanced search, sorted the results based on relevance, and screened the first 100 search results for relevant trials. For details of our search strategies, please refer to the appendices (Appendix 2).

Searching other resources

Reference lists
We checked the bibliography of the included trial for further references to relevant trials. We also searched research papers which cannot be directly included in this review (observational studies, systematic and narrative reviews, conference reports, etc.) for the citations of relevant trials which may have been eligible for inclusion. These papers were then sought and assessed for possible inclusion in the review.

Authors and organisations
We contacted the trial authors who have conducted prominent research in the relevant field regarding their ongoing trials or other relevant papers which may be eligible for inclusion. We contacted the following organisations for further information:

• NCCIH formerly known as NCCAM;

• Foundation for Alternative and Integrative Medicine;

• American Alternative Medical Association;

• The Complementary and Natural Healthcare Council;

• International Alternative Medical Association;

• The Association for Applied Psychophysiology and Biofeedback (formerly the Biofeedback Society of America).

Grey literature
Regarding grey literature, research has shown that published studies often overestimate outcomes, compared to grey literature articles (Hopewell 2007). Moreover, exclusion of grey literature can result in systematic error, thus seriously threatening the validity of a systematic review (McAuley 2000). Therefore, to avoid this risk of publication bias, we searched grey literature databases in our attempt to identify relevant trials and authors from conference proceedings. We searched the following databases (recommended in the Cochrane Handbook For Systematic Reviews of Interventions) for unpublished reports or articles which may be appropriate for inclusion in this systematic review (Higgins 2011a):

• OpenGrey (www.opengrey.eu/: searched on 04 September 2017);

• PsychEXTRA (www.apa.org/psycextra/; searched on 04 September 2017);

• the Grey Literature Report (http://greylit.org/; searched on 04 September 2017);

• the Agency for Healthcare Research and Quality (www.ahrq.gov/; searched on 04 September 2017);

• MedNar (http://mednar.com/mednar/desktop/en/search.html; searched on 04 September 2017).

For details of our search strategies, please refer to the appendices (Appendix 2).

Selection of studies

Two review authors (SP, MG) independently assessed the eligibility of the trials identified by the literature searches by completing a trial selection form that was designed in accordance with the inclusion criteria. Each author independently evaluated each title for inclusion. If we did not find the relevant information in the abstract, we tried to retrieve the relevant full text, report(s). Both authors approved the inclusion of the trial in the review. In the case of any discrepancy related to eligibility, a third author (SKB) would have arbitrated, however, no such discrepancy occurred. We tabulated the excluded trials under 'Characteristics of excluded studies' and gave reasons for the exclusion. The review authors were not blinded to the identity of the trial authors, institutions or trial results during their assessments. We also produced a PRISMA chart to illustrate the trial selection process (Figure 1).

Figure 1

---

---

Study flow diagram.

Data extraction and management

Two review authors (SP, SM) independently and simultaneously extracted data from the selected articles and this was verified by two other review authors with an aim to resolve any disagreements by discussion and consensus.

We extracted data using a standardised, pre‐tested data extraction form. The review authors designed the data extraction form together and pilot tested it using a sample RCT and a quasi‐RCT to ensure practical functionality.

We extracted the following information from the included trial.

Assessment of risk of bias in included studies

Two review authors (RD, SM) independently assessed the risk of bias of the included trial by using the criteria outlined in chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b).

We assessed each trial according to the following six components.

1. Random sequence generation (selection bias)

2. Allocation concealment (selection bias)

3. Masking (blinding) of participants and personnel (performance bias), and masking of outcome assessor.

4. Incomplete outcome data (attrition bias) through withdrawals, dropouts and protocol deviations; and

5. Selective reporting bias

6. Other bias

We also assessed for any other sources of bias as reported in the Cochrane Handbook for Systematic Reviews of Interventions (bias related to the specific trial design, early or abrupt end of trials, fraudulent trials) (Higgins 2011b). For each of the mentioned components, we assigned judgements of either low, unclear or high risk of bias.

We recorded the results in the standard table provided in the Review Manager software (RevMan 2014), and summarised the findings in a 'Risk of bias' table or graph. We resolved all concerns or issues by discussion with a third review author (ALA). We also addressed the implications of the 'Risk of bias' assessment in the discussion section.

Measures of treatment effect

For future updates of this review, if we include more trials, we plan to analyse and present dichotomous data (e.g. improvement in pain rating, analgesic usage, and overall satisfaction with TENS) using the risk ratios (RR) with 95% confidence intervals (CIs). We will report adverse effects of the intervention with 99% CIs to avoid errors due to multiple statistical testing. We will analyse continuous data (e.g. pain scores) using the mean difference (MD) and 95% CIs. We will also calculate the standardised MD (SMD) if different scales of measurement have been used.

We will present count data (e.g. frequency of pain episodes) as the MD, by comparing the difference in the mean number of events in the intervention group as compared to the control group and present them as rate ratio with 95% CI in case of rare events.

If we identify trials with multiple treatment arms, we will only include those treatment arms whose parameters are minimally different from the other included trials.

In this current version of the review, since no meta‐analyses were possible, we have presented a narrative summary.

Unit of analysis issues

We planned to use individuals with SCD as the unit of analysis in this review.

Regarding cross‐over trials, as we are currently unaware of the long‐term effects of these interventions, we are unable to determine whether the effects of the first intervention will interfere with those of the second. In order to avoid introducing this bias into the analysis, we aimed only to include first‐arm data from cross‐over trials, when available. The included trial did not present first‐arm data and we were unable to analyse any data from this trial, also whilst reporting on 22 participants, data on 60 treatment episodes were presented (30 for each treatment group). We have therefore only reported on the included trial in a narrative way.

For future updates, we intend to report any included cluster‐randomised trials separately.

Dealing with missing data

We contacted the trial authors and requested access to the missing data regarding methods, participants, interventions and outcomes. The lead investigator on the included trial provided the original protocol (Wang 1988).

Assessment of heterogeneity

For future updates, if more trials are included in the review, in order to deal with clinical heterogeneity, we will pool trials that examine similar interventions. We will perform separate analyses for TENS compared to conventional treatment and TENS compared to placebo TENS or other non‐pharmacological treatment. We will use the Chi² test for assessing heterogeneity (significance level: P < 0.1). We will quantify the degree of heterogeneity using the I² statistic (Deeks 2008). The guidelines for interpretation of the I² values will be as follows.

• 0% to 40% indicates unimportant levels of heterogeneity

• 30% to 60% indicates moderate heterogeneity

• 50% to 90% indicates substantial heterogeneity

• 75% to 100% indicates considerable heterogeneity

We will also consider a visual inspection of the forest plot to identify whether CIs overlap.

Assessment of reporting biases

For the included trial, we compared the protocol provided by the author with the published report (Wang 1988). We also requested and obtained the raw data for comparison with the published data. However, the raw data did not provide us with either the baseline pain score or the SD. In the absence of any other eligible trial, we could not impute the SD, hence we were unable to conduct any paired analysis.

Although this version of the review does not include trials from the grey literature, we do anticipate that these may be included in future updates. Therefore, it is particularly important for us to conduct an assessment of publication bias. For future updates, if we identify at least 10 trials for inclusion in a meta‐analysis, we will explore potential publication bias by generating a funnel plot and performing Egger’s test to determine the degree of asymmetry (Egger 1997). If the included trials differ in sample size then we will visually inspect the funnel plots to explore the possibility of reporting biases. We will interpret the results of Egger’s test cautiously due to the presence of many limitations in quantifying possible reporting biases (Moore 2008). We will address the implications of the publication bias assessment on the review findings in the discussion section.

Data synthesis

We synthesised the quantitative data compiled using the data extraction form but did not undertake any meta‐analyses due to issues of the trial design and reporting. We will calculate the I² statistic to determine statistical heterogeneity, and thus determine whether we will use a fixed‐effect model (unimportant heterogeneity), or a random‐effects model (at least moderate heterogeneity) for the meta‐analysis (DerSimonian 1986). The results of the meta‐analysis will be reported as pooled effect estimates, stated with 95% CIs and illustrated graphically in a forest plot.

Subgroup analysis and investigation of heterogeneity

In future, if we identify substantial (or higher) statistical heterogeneity we plan the following subgroup analysis investigating the possible effect of TENS based on:

• age groups (below 18 years of age, 18 years and over);

• gender (male, female);

• stimulation parameters (e.g. high stimulation frequency with low intensity, low frequency with a high stimulus intensity);

• frequency of administration of TENS (e.g. daily or not on a daily basis);

• duration of treatment with TENS (e.g. 30‐minute sessions and under or more than 30 minutes).

Sensitivity analysis

If appropriate we will assess the robustness of our findings by performing sensitivity analyses according to Cochrane recommendations (Deeks 2011). If there are sufficient comparable trials, i.e. 10 or more, we will perform sensitivity analyses to study the effect of excluding trials with high risks of bias due to inadequate allocation of concealment, blinding, randomisation method or level of dropouts. Furthermore, we will also consider the impact of using a fixed‐effect model compared to a random‐effects model.