Electrical stimulation for treating pressure ulcers

Mohit Arora; Lisa A Harvey; Joanne V Glinsky; Lianne Nier; Lucija Lavrencic; Annette Kifley; Ian D Cameron

doi:10.1002/14651858.CD012196.pub2

Electrical stimulation for treating pressure ulcers

Authors' declarations of interest

Version published: 22 January 2020 Version history

https://doi.org/10.1002/14651858.CD012196.pub2

Collapse all Expand all

Abstract

available in

Background

Pressure ulcers (also known as pressure sores, decubitus ulcers or bedsores) are localised injuries to the skin or underlying tissue, or both. Pressure ulcers are a disabling consequence of immobility. Electrical stimulation (ES) is widely used for the treatment of pressure ulcers. However, it is not clear whether ES is effective.

Objectives

To determine the effects (benefits and harms) of electrical stimulation (ES) for treating pressure ulcers.

Search methods

In July 2019 we searched the Cochrane Wounds Specialised Register; the Cochrane Central Register of Controlled Trials (CENTRAL); Ovid MEDLINE (including In‐Process & Other Non‐Indexed Citations); Ovid Embase and EBSCO CINAHL Plus. We also searched clinical trials registries for ongoing and unpublished studies, and scanned reference lists of relevant included studies as well as reviews, meta‐analyses and health technology reports to identify additional studies. We did not impose any restrictions with respect to language, date of publication or study setting.

Selection criteria

We included published and unpublished randomised controlled trials (RCTs) comparing ES (plus standard care) with sham/no ES (plus standard care) for treating pressure ulcers.

Data collection and analysis

Two review authors independently selected trials for inclusion, extracted data, and assessed risk of bias. We assessed the certainty of evidence using GRADE.

Main results

We included 20 studies with 913 participants. The mean age of participants ranged from 26 to 83 years; 50% were male. ES was administered for a median (interquartile range (IQR)) duration of five (4 to 8) hours per week. The chronicity of the pressure ulcers was variable, ranging from a mean of four days to more than 12 months. Most of the pressure ulcers were on the sacral and coccygeal region (30%), and most were stage III (45%). Half the studies were at risk of performance and detection bias, and 25% were at risk of attrition and selective reporting bias. Overall, the GRADE assessment of the certainty of evidence for outcomes was moderate to very low. Nineteen studies were conducted in four different settings, including rehabilitation and geriatric hospitals, medical centres, a residential care centre, and a community‐based centre.

ES probably increases the proportion of pressure ulcers healed compared with no ES (risk ratio (RR) 1.99, 95% confidence interval (CI) 1.39 to 2.85; I² = 0%; 11 studies, 501 participants (512 pressure ulcers)). We downgraded the evidence to moderate certainty due to risk of bias.

It is uncertain whether ES decreases pressure ulcer severity on a composite measure compared with no ES (mean difference (MD) ‐2.43, 95% CI ‐6.14 to 1.28; 1 study, 15 participants (15 pressure ulcers) and whether ES decreases the surface area of pressure ulcers when compared with no ES (12 studies; 494 participants (505 pressure ulcers)). Data for the surface area of pressure ulcers were not pooled because there was considerable statistical heterogeneity between studies (I² = 96%) but the point estimates for the MD of each study ranged from ‐0.90 cm² to 10.37 cm². We downgraded the evidence to very low certainty due to risk of bias, inconsistency and imprecision.

It is uncertain whether ES decreases the time to complete healing of pressure ulcers compared with no ES (hazard ratio (HR) 1.06, 95% CI 0.47 to 2.41; I² = 0%; 2 studies, 55 participants (55 pressure ulcers)). We downgraded the evidence to very low certainty due to risk of bias, indirectness and imprecision.

ES may be associated with an excess of, or difference in, adverse events (13 studies; 586 participants (602 pressure ulcers)). Data for adverse events were not pooled but the types of reported adverse events included skin redness, itchy skin, dizziness and delusions, deterioration of the pressure ulcer, limb amputation, and occasionally death. We downgraded the evidence to low certainty due to risk of selection and attrition bias and imprecision.

ES probably increases the rate of pressure ulcer healing compared with no ES (MD 4.59% per week, 95% CI 3.49 to 5.69; I² = 25%; 12 studies, 561 participants (613 pressure ulcers)). We downgraded the evidence to moderate certainty due to risk of bias. We did not find any studies that looked at quality of life, depression, or consumers' perception of treatment effectiveness.

Authors' conclusions

ES probably increases the proportion of pressure ulcers healed and the rate of pressure ulcer healing (moderate certainty evidence), but its effect on time to complete healing is uncertain compared with no ES (very low certainty evidence). It is also uncertain whether ES decreases the surface area of pressure ulcers. The evidence to date is insufficient to support the widespread use of ES for pressure ulcers outside of research. Future research needs to focus on large‐scale trials to determine the effect of ES on all key outcomes.

PICOs

Population

Intervention

Comparison

Outcome

The PICO model is widely used and taught in evidence-based health care as a strategy for formulating questions and search strategies and for characterizing clinical studies or meta-analyses. PICO stands for four different potential components of a clinical question: Patient, Population or Problem; Intervention; Comparison; Outcome.

See more on using PICO in the Cochrane Handbook.

Plain language summary

available in

Is electrical stimulation effective for treating pressure ulcers?

What is the aim of this review?

The aim of this review was to find out whether electrical stimulation (ES; an electrical current applied to the skin) can help heal pressure ulcers. We collected and analysed all relevant studies (randomised controlled trials) to answer this question and found 20 relevant studies.

Key messages

ES compared with no ES probably increases the proportion of pressure ulcers healed and the rate of pressure ulcer healing (moderate certainty evidence) but its effect on time to complete healing and the surface area of pressure ulcers is uncertain (very low certainty evidence). The most commonly reported side effects of ES were reddening of the skin and discomfort. There is a need for better quality research to determine whether ES is safe and effective.

What was studied in the review?

Pressure ulcers (also known as pressure sores, bed sores or pressure injuries) are injuries to the skin and/or underlying tissue caused by sustained pressure over bony parts of the body such as the hips, heels or lower back. People with reduced mobility due to age, disability or illness are at risk of developing pressure ulcers.

ES is provided by an electrical current that can be applied to the skin in different ways. ES requires the placing of at least two small electrodes on the skin connected to a small battery‐powered device which controls the intensity of the current. ES can be delivered either as a direct or pulsed current. It causes a tingling or vibratory sensation in most people except those who cannot feel due to conditions such as spinal cord injury. We reviewed the evidence about whether ES affects the number of pressure ulcers healed, the size and severity of the pressure ulcers, the time to complete healing, and quality of life. We also wanted to find out about any side effects associated with ES.

What are the main results of the review?

This review includes the results of 20 randomised controlled trials dating from 1985 to 2018 and involving 913 participants. The average age of participants ranged from 26 to 83 years; 50% were male. Participants had their pressure ulcers for at least four days and in some cases for more than 12 months. The majority of pressure ulcers (60%) were serious and on or adjacent to the buttocks (62%). Studies were conducted in four different settings, including rehabilitation and geriatric hospitals, medical centres, a residential care centre, and a community‐based centre. ES was administered for an average of five hours per week. Studies compared ES plus usual care (e.g. wound dressing, pressure relief, regular turning, nutritional advice and supplements) to no ES (but with usual care). Eight studies out of 20 were funded by a device manufacturer with a vested interest in the results of the studies.

Eleven studies that compared ES with no ES indicated that ES probably improves the proportion of pressure ulcers healed (moderate certainty evidence based on 501 participants (512 pressure ulcers)). It is uncertain whether ES decreases pressure ulcer severity on a composite measure (based on 1 study with 15 participants (15 pressure ulcers)). The effect of ES on pressure ulcer area was not estimable because different studies showed very different results. It is uncertain whether ES decreases the surface area of pressure ulcers (very low certainty evidence based on 494 participants (505 pressure ulcers)). We cannot be certain whether ES has an effect on time to complete healing (very low certainty evidence based on 55 participants (55 pressure ulcers)). The common complications related to ES were skin redness and discomfort (low certainty evidence based on 586 participants (602 pressure ulcers)). Twelve studies also indicated that ES probably increases the rate of pressure ulcer healing (moderate certainty evidence based on 561 participants (613 pressure ulcers)). No studies reported results for quality of life or depression.

How up‐to‐date is this review?

We searched for studies that had been published up to July 2019.

Authors' conclusions

available in

Implications for practice

Electrical stimulation (ES) probably increases the proportion of pressure ulcers healed, but its effect on time to complete healing is uncertain, and the certainty of evidence for all outcomes is moderate, low or very low. The evidence to date is insufficient to support the widespread use of ES for pressure ulcers other than for research purposes.

Implications for research

We have two broad recommendations for future research in this area. One is about research design and reporting, and the other is about areas of future research.

Our recommendations about research design and reporting are as follows.

Future studies must focus on minimising bias. Importantly, they need to randomly allocate participants to groups, use concealed allocation and blind assessors. Adequate follow‐up of participants will also help minimise bias. Importantly, studies need to register their protocols and adhere to preplanned statistical analyses.
Studies must report between‐group differences for all outcomes. They also need to examine both the short‐term (i.e. within 4 weeks of the cessation of treatment) and long‐term effects of ES (i.e. more than 4 weeks after the cessation of treatment). This will aid interpretation of results and future meta‐analyses.
Studies must follow CONSORT reporting guidelines (Moher 2010). Most studies included in this review only adhered to a few items of CONSORT. Full adherence will have an important positive impact on research in this area.
Studies must pay particular attention to some key methodological issues. In addition, time to healing data need to be appropriately analysed using Kaplan‐Meier curves and hazard ratios. The area would also benefit from consensus on key outcomes and endpoints for future studies.
Future studies should ensure they have a sufficient sample size to detect clinically important differences.
Trialists should take care to avoid unit of analyses problems by randomising, reporting and analysing at the level of the participant not the pressure ulcer.

Our recommendations about areas of future research are as follows.

Studies need to determine the effect of ES on all key outcomes. There is a pressing need to include outcome measures that are important to patients. This needs to include time to complete healing, quality of life, depression and consumers' perceptions of treatments. Future studies should also include composite measures, reflecting the severity of pressure ulcers. The PUSH tool would seem particularly appropriate given it is widely recommended in various guidelines including the NPUAP (NPUAP/EPUAP 2014).
Studies need to determine whether ES has any adverse events. No study included in this review adequately looked at this issue, yet it is feasible that ES could be harmful. This in an important issue that needs attention.
Studies need to look at the long‐term effects of ES. Only one study included in this review looked at the long‐term effects of ES (Adunksy 2005), yet, the long‐term effects are perhaps of most interest and relevance to people with pressure ulcers.
Studies need to look at the cost‐effectiveness of ES for the treatment of pressure ulcers. This should be an important consideration before its widespread use.
Studies need to determine the treatment burden for people with pressure ulcers. A better understanding of this would help guide decisions about minimally worthwhile treatment effects.
Studies need to focus on the optimal therapeutic frequency, duration, and location of treatment. There is no consensus on these parameters, and they are important issues to be considered in future studies.
Studies need to explore the possible mechanism of action of ES. That is, determine how ES could heal pressure ulcers. This will guide decisions about the optimal treatment parameters and electrode placement.

Summary of findings

Open in table viewer

Summary of findings for the main comparison. Electrical stimulation (plus standard care) versus sham/no ES (plus standard care) for treating pressure ulcers

Electrical stimulation (plus standard care) versus sham/no ES (plus standard care) for treating pressure ulcers
Patient or population: people with pressure ulcers Setting: inpatients and outpatients Intervention: electrical stimulation (plus standard care) Comparison: sham/no ES (plus standard care)
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	Number of ulcers (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with sham/no ES (plus standard care)	Risk with Electrical stimulation (plus standard care)
Proportion of pressure ulcers healed (3 to 12 weeks)	Study population		RR 1.99 (1.39 to 2.85)	512 (11 RCTs)	⊕⊕⊕⊝ Moderate^a	ES may increase the proportion of pressure ulcers healed when compared with no ES. Absolute effect: 297 out of 1000 (from 207 more to 425 more).
	149 per 1,000	297 per 1,000 (207 to 425)
Time to complete healing (3 and 8 weeks)	Study population		HR 1.06 (0.47 to 2.41)	55 (2 RCTs)	⊕⊝⊝⊝ Very low^b	It is uncertain if ES decreases time to complete healing when compared with no ES.
	18 per 100	19 per 100 (9 to 38)
Complications/ adverse events related to pressure ulcers (3 to 12 weeks)	Adverse events included redness of the skin, itchy skin, dizziness and delusions, deterioration of the pressure ulcer, limb amputation and occasionally death.			602 (13 RCTs)	⊕⊕⊝⊝ Low^c	The data were not sufficiently detailed or comparable to analyse quantitatively.
Quality of life (QoL)	No studies measured quality of life
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RR: Risk ratio; OR: Odds ratio;
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: we have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect
^aDowngraded one level: once for serious risk of bias because a lot of the studies had either high or unclear risk of bias for performance bias and selective reporting. ^bDowngraded four levels: once for serious risk of bias because both studies had high risk of bias for two domains and one study had unclear risk of bias for another three domains; once for indirectness because the two studies were not reflective of all who are vulnerable to pressure ulcers; twice for imprecision. ^cDowngraded two levels: once for serious risk of bias because a lot of the studies had either high or unclear risk of bias for selection and attrition bias; once for imprecision.

Background

available in

Description of the condition

Pressure ulcers (also known as pressure sores, pressure injuries, decubitus ulcers or bedsores) are localised injuries to the skin or underlying tissue, or both. Pressure ulcers usually occur over a bony prominence and are caused by pressure, friction or shear forces. Friction or shear forces occur when layers of the skin are forced to slide over one another or over deeper layers of tissue (NPUAP/EPUAP 2014).

Pressure ulcers are a disabling consequence of immobility. They most often occur in people with neurological conditions (e.g. people with spinal cord injuries (Rintala 2008), acute illnesses (for example, people in comas in intensive care units) (Schoonhoven 2006), or in people who are elderly and immobile (for example, older people in nursing home care) (Perneger 2002).

Pressure ulcers are a common problem (Cowan 2019). They affect up to 32% of people admitted to hospital (Kaltenthaler 2001), and 40% of people with spinal cord injuries (Zakrasek 2015). They hinder rehabilitation and have many harmful consequences. For example, they can lead to contractures (shortening of muscles, tendons, or ligaments), permanent scarring, deformities, osteomyelitis (infection of bones), loss of limbs, and sepsis (a life‐threatening response to infection) (Allman 1989; Rodriguez 1994). People with severe pressure ulcers commonly require hospitalisation. Pressure ulcers affect health‐related quality of life and participation in meaningful community activities (New 2004). In addition, they affect a person's family life, and are costly and difficult to manage (Brem 2010). Pressure ulcers can also be life‐threatening, particularly in low‐ and middle‐income countries (Hossain 2015; Zakrasek 2015).

Two key guidelines recommend using the Pressure Ulcer Classification System to classify the severity of pressure ulcers (NPUAP/EPUAP 2014). This system is based on the level of tissue injury, and classifies pressure ulcers into four 'stages' ('grades' or 'categories') with two additional unstageable categories. A stage I pressure ulcer is indicated by non‐blanching superficial red areas, and a stage IV pressure ulcer is indicated by full skin thickness injuries that can involve the underlying bone, tendon or joint capsule, and invariably requires hospitalisation and surgery (see Appendix 1 for further details of the Pressure Ulcer Classification System).

Description of the intervention

Electrical stimulation (ES) is advocated as a way of healing pressure ulcers (Bogie 2000), and is recommended in at least four clinical practice guidelines (AWMA 2012; Consortium for Spinal Cord Medicine 2014; Houghton 2013; SCIRE 2014). It is provided by an electrical current that can be applied in different ways, however, in this review we are only investigating ES that is applied on the skin. Application requires placing at least two electrodes on the skin, which are connected to a small battery‐like device. The intensity of the ES is controlled through dials or switches. The cost of an ES device ranges from USD 80 to USD 750 (Mittmann 2011).

ES for the treatment of pressure ulcers can be delivered either as a direct or pulsed current. When a direct current is used, the current flows constantly in one direction. When a pulsed current is used, each pulse is separated by a period of no flow of current. There are two types of pulsed current; monophasic and biphasic. In both types of pulsed current, the electric current is delivered in short bursts, however in monophasic, the current flows in one direction, whilst in biphasic the current flows in two directions.

There are different ways of placing the electrodes, for example, they can be placed in or around the pressure ulcer, or on other parts of the body. In addition, the parameters of the ES can be varied. This includes the frequency (low or high Hz), polarity (negative, positive or mixed), pulse type (monophasic or biphasic), duration of stimulation (per session) and amplitude (low or high mA). ES causes a comfortable tingling or vibratory sensation (except in those with neurological lesions that results in the loss of sensation) and can cause a muscle contraction.

How the intervention might work

There are many theories about how ES may help heal pressure ulcers but the veracity of these theories is unclear. Most work in this area was done in the 1980s with very little recent work directed at furthering our understanding. Researchers have suggested that ES affects all four phases of healing, that is, the inflammatory, proliferative, epithelialisation and remodelling phases. Most believe that ES increases blood flow to the affected area (Alvarez 1983; Bourguignon 1987; Cruz 1989). This may increase the flow of cells important for the inflammatory and proliferation phases (Foulds 1983; Orida 1982), or promote tissue oxygenation and reduce oedema (Sussman 2012). It may also influence the increase of epidermal growth factors and their receptors (Zhao 2002). Some have even suggested that ES has an antibacterial effect that helps reduce infection and enhance healing (Fakhri 1987). However, none of these theories have been substantiated and the effects of ES have not been reviewed systematically.

In addition, ES may indirectly help treat and prevent pressure ulcers in people with neurological disorders by its possible effects on the properties of muscles. For example, some clinicians apply ES to facilitate a contraction of the gluteal muscles of people with paralysis. The ES is used to induce muscle hypertrophy and hence better distribute pressure over the ischial tuberosities; a region that is highly vulnerable to pressure ulcers with prolonged sitting. However, there is no strong evidence to support these beliefs.

Why it is important to do this review

This review is important because pressure ulcers are very common and debilitating, and there is initial evidence to suggest that ES is therapeutic. There are a small number of non‐Cochrane reviews that claim that ES is effective for the treatment of pressure ulcers (Barnes 2014; Kawasaki 2014; Lala 2016; Liu 2014). These have prompted clinical guidelines to start recommending ES (AWMA 2012; Houghton 2013; NPUAP/EPUAP 2014), however, it is not clear whether these recommendations are justified because the reviews and studies that they are based upon have methodological limitations. It is important to know the certainty of the evidence that underpins any recommendation for ES because ES is costly, time‐consuming to administer, and inconvenient for patients. It also requires specialised equipment, training and daily application. In addition, there is the potential for harm. For example, ES can cause electric burns and it is possible that ES could hinder the healing of pressure ulcers. Therefore it is important to establish whether ES is effective, whether the potential for therapeutic effect outweighs any potential for harm, and whether the associated cost, time and inconvenience of ES are justified.

Objectives

available in

To determine the effects (benefits and harms) of electrical stimulation (ES) for treating pressure ulcers.

Methods

available in

Criteria for considering studies for this review

Types of studies

We included published and unpublished randomised controlled trials (RCTs). We included studies irrespective of language of publication. We included studies that used parallel‐group designs, within‐participant designs or cross‐over designs.

Types of participants

We included participants of all ages and both genders, provided participants had at least one pressure ulcer. We excluded studies that only involved participants with other types of wounds (e.g. diabetic and venous ulcers). If a study involved participants with different types of wounds, we extracted the data for participants with pressure ulcers. If this was not possible, we only included the study if more than 75% of participants had pressure ulcers. There were no restrictions on the type or stage of the pressure ulcers, that is, we included acute or chronic pressure ulcers of any stage (including non‐open wounds classified as stage I) and due to any cause. We anticipated most pressure ulcers would be due to neurological conditions (e.g. people with spinal cord injuries) and acute illnesses (for example, people in comas in intensive care units), or due to age and immobility (e.g. older people in nursing homes). We included studies even if the causes of the pressure ulcers were not reported but it was reasonable to assume that they were due to pressure injuries (e.g. the pressure ulcers were on the sacrum).

Types of interventions

We included studies that determined the effectiveness of any type of ES for treating pressure ulcers. We included studies that compared ES (plus standard care) with sham/no ES (plus standard care).

We included ES which was administered through either direct or pulsed current. Standard care could include any of the following: wound dressing, pressure relief, regular turning, nutritional advice, and nutritional supplements.

Types of outcome measures

Our primary outcomes are mainly reflective of pressure ulcer healing. For example, proportion of pressure ulcers healed, composite measures of pressure ulcer severity, surface area of pressure ulcers and time to complete healing. This focus is justified because ES for pressure ulcers is primarily administered in an effort to promote healing.

Primary outcomes

Proportion of pressure ulcers healed; the data expressed as the number of pressure ulcers healed in each group.
Composite measures of pressure ulcers that captured different aspects of severity; this includes measures such as the Pressure Ulcer Scale for Healing (PUSH) (Gardner 2005), Sussman Wound Healing Tool (Sussman 1997) and Pressure Sore Status Tool (Bates‐Jensen 1992).
Surface area of pressure ulcers; data expressed as cm². If not provided in the study, areas were calculated by multiplying the length and width of the pressure ulcers.
Time to complete healing; these data expressed as days to wound closure (time‐to‐event data).
Complications/adverse events; these include death, skin irritation, spasm, or pain, or the number of pressure ulcer infections.

Secondary outcomes

Rate of pressure ulcer healing expressed as percentage rate of healing per week.
Quality of life; including any validated standardised questionnaire that captured quality of life. For example, the Short Form‐36 (Forchheimer 2004) and Euro Quality of Life (Whitehurst 2012).
Depression; this includes any validated standardised outcome that captures depression e.g., the Hospital and Anxiety Depression Scale (Woolrich 2006) and Depression Anxiety Stress Scale (Üstün 2010).
Consumers' perception of treatment effectiveness; this includes any outcome that captured consumers' satisfaction, impression of treatment effectiveness or comfort with ES.

We only extracted one type of measure from a study to reflect each of the primary and secondary outcomes. If a study had more than one type of measure for any primary or secondary outcome, then we chose the measure that was most valid and reliable.

Search methods for identification of studies

Electronic searches

We searched the following electronic databases to identify reports of relevant clinical trials:

the Cochrane Wounds Specialised Register (searched 02 July 2019);
the Cochrane Central Register of Controlled Trials (CENTRAL; 2019, Issue 6) in the Cochrane Library (searched 02 July 2019);
Ovid MEDLINE including In‐Process & Other Non‐Indexed Citations (1946 to 02 July 2019);
Ovid Embase (1974 to 02 July 2019);
EBSCO CINAHL Plus (Cumulative Index to Nursing and Allied Health Literature; 1937 to 02 July 2019);
PEDro (www.pedro.org.au) (Physiotherapy Evidence Database; 1929 to 02 July 2019).

The search strategies for the Cochrane Wounds Specialised Register, CENTRAL, Ovid MEDLINE, Ovid Embase and EBSCO CINAHL Plus and PEDro can be found in Appendix 2. We combined the Ovid MEDLINE search with the Cochrane Highly Sensitive Search Strategy for identifying randomised trials in MEDLINE: sensitivity‐ and precision‐maximising version (2008 revision) Lefebvre 2019). We combined the Embase search with the Ovid Embase filter developed by the UK Cochrane Centre (Lefebvre 2019). We combined the CINAHL Plus searches with the trial filters developed by the Scottish Intercollegiate Guidelines Network (SIGN 2018). We did not impose any restrictions with respect to language, date of publication or study setting.

We also searched the following clinical trials registries to identify unpublished and ongoing studies (searched 02 July 2019) (see Appendix 2 for search terms).

WHO International Clinical Trials Registry Platform (ICTRP) (www.who.int/trialsearch);
ClinicalTrials.gov (www.clinicaltrials.gov);
the International Standard Randomised Controlled Trial Number (ISRCTN) registry (www.controlled‐trials.com);
EU Clinical Trials Register (www.clinicaltrialsregister.eu);
Australian New Zealand Clinical Trials Registry (www.anzctr.org.au);
Stroke Trials Registry (www.strokecenter.org/trials).

Searching other resources

To identify further published, unpublished and ongoing studies, we:

used the Cited Reference Search within Web of Science (Thomson Reuters) Science Citation Index (SCI) and Social Science Citation Index (SSCI) (searched 02 July 2019) to track relevant references (see Appendix 2 for search terms);
scanned the reference lists of all identified studies and reviews;
searched grey literature using Open Grey (www.opengrey.eu), Google Scholar (scholar.google.com), and Proquest Dissertations & Theses databases;
contacted key researchers in the area and international organisations to enquire about unpublished or ongoing studies;
contacted manufacturers of ES devices and authors regarding any published or unpublished data.

Searching reference lists of included trials and relevant reviews

We aimed to identify other potentially eligible trials or ancillary publications by searching the reference lists of retrieved included trials, as well as relevant systematic reviews, meta‐analyses and health technology assessment reports.

Searching by contacting individuals or organisations

When necessary, we contacted authors of key papers and abstracts to request further information about their trials.

Adverse effects

We did not perform a separate search for adverse effects of interventions used, we considered adverse effects described in included studies only.

Data collection and analysis

We carried out data collection and analysis according to methods stated in the published protocol (Arora 2016), which were based on the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a).

Selection of studies

Two review authors (MA and LAH) independently screened the titles and abstracts of the search output to identify potentially relevant studies. We retrieved full‐length articles for all potentially relevant studies and used these to identify studies that met the selection criteria. The full‐length articles were re‐examined to ensure that they met the inclusion criteria. Review authors did not screen studies in which they were involved. In such instances studies were screened by another review author who was not involved in the study. Disagreements between the two review authors (MA and LAH) were resolved by discussion and, when necessary, arbitrated by a third review author (JVG).

We compiled a table of the excluded studies and detailed the primary reason for exclusion. We created a flow diagram using the PRISMA template within Review Manager 2014 (Liberati 2009). The flowchart included the number of:

records identified by the database and other searches;
records after removal of duplicates;
records excluded after preliminary screening (i.e. of titles and abstracts);
records retrieved in full text;
records or studies excluded after assessment of the full text with brief reasons;
studies included in qualitative synthesis and quantitative syntheses (meta‐analysis).

Data extraction and management

Two review authors (MA and LAH) independently performed data extraction for all included studies. Differences between the two review authors were resolved by discussion and, when necessary, arbitrated by a third author (JVG). If data were missing from reports, we attempted to contact the study authors to obtain the missing data. We resolved discrepancies by consensus.

We extracted the data from included studies into an Excel spreadsheet designed to capture the trial information detailed below. Initially, we piloted the Excel spreadsheet to explore any issues that may arise in relation to the data extraction process. We expanded and amended the spreadsheet as necessary after the piloting process.

We extracted the maximal amount of data without duplicating results from dual publications. We extracted the following data as listed below.

Author: year of publication
Methods: study design
Participants: health condition; sample size; study setting and country; inclusion and exclusion criteria; characteristics of pressure ulcers; age; gender
Interventions: details of the experimental group (i.e. duration of ES, electrode placement, name of device and manufacturer, intensity of ES, type of current and polarity) and control group; details of co interventions
Outcomes: details of outcomes included in the review; other outcomes not included in the review; time points (i.e. when outcomes were measured)
Withdrawals and reason for withdrawals
Funding source; registry; published protocol

Assessment of risk of bias in included studies

Two review authors (MA and LAH) assessed the risk of bias in each study using the following eight methodological domains as recommended by the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b).

Random sequence generation
Allocation sequence concealment
Blinding of participants
Blinding of personnel
Blinding of outcome assessors
Incomplete outcome data
Selective outcome reporting
Other potential sources of bias

We assessed each of the eight domains for low, high, or unclear risk of bias (see Appendix 3). We rated studies as high risk of bias for incomplete outcome data if more than 15% of participants had dropped out of the study (this decision was made after the protocol was published).

We attempted to contact the study authors, wherever applicable, to clarify any ambiguities. Disagreements in judgements about the risk of bias were resolved by discussion or, when necessary, arbitrated by an independent third review author (JVG). Review authors did not extract data, or rate the risk of bias of studies in which they were involved. In such instances these tasks were performed by two authors who were not involved in conducting the study.

We used Review Manager 5 to generate two figures detailing the risk of bias (Review Manager 2014). The first figure (the 'Risk of bias' graph) was used to illustrate the judgements about the risk of bias ('low risk', 'high risk', 'unclear risk' of bias) for each study. The second figure (the 'Risk of bias' summary) was used to present the judgements about the risk of bias in a cross‐tabulation format.

Measures of treatment effect

For continuous data:

we expressed mean differences (MDs) with 95% confidence intervals (CIs) for outcomes that used the same units (this was done for surface area of pressure ulcers and rate of pressure ulcer healing);
we planned to express summary estimates as standardised MDs (SMDs) with 95% CIs for outcomes that used different units.

We converted available data, where possible, using the calculator incorporated into Review Manager 5 (e.g. when data were reported as standard errors) (Higgins 2011a; Review Manager 2014). Change scores were given preference over postintervention scores, however, we did not intend to combine postintervention scores with change scores in meta‐analyses using SMDs, as suggested by the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2011). Ultimately, this was not relevant because we did not express any results as SMDs.

Wherever possible, data measured on the same scale but with different units were converted to the same units (this was done for surface area of pressure ulcers and rate of pressure ulcer healing). In addition, we planned to express rate of pressure ulcer healing as either mm² healed per day, cm² healed per day or percentage healed per day. However, ultimately, we expressed these data as percentage healed per week because this was how most authors presented these data. This required converting some data that were expressed differently by an appropriate conversion constant (e.g. percentage healed per day was multiplied by 7 and percentage healed per 4 weeks was divided by 4 to obtain percentage healed per week).

We planned to convert SMD from meta‐analyses into MD to aid clinical interpretation (Deeks 2011). We intended to do this by using an outcome and its standard difference (SD) from one of the studies included in the meta‐analysis. We planned to choose an outcome that was widely used, provided the study from which it was taken had a reasonable sample size. We planned to calculate the MD by multiplying the SMD by the baseline SD from the control group of the selected study. Ultimately, this was not done because we did not express any results as SMDs.

For dichotomous data, we expressed summary estimates as:

risk ratios (RRs) with 95% CIs (this was done for proportion of pressure ulcers healed).

For time‐to‐event data, we expressed summary estimates as:

hazard ratios (HRs) with 95% CIs (this was done for time to complete healing).

Unit of analysis issues

We planned to include cross‐over studies, studies with more than one ES group, studies in which multiple observations were taken on the same individual, and studies in which more than one pressure ulcer was treated per participant. Where these types of studies were included in the review we dealt with them in the following way.

Cross‐over studies

We planned to use the first period of cross‐over studies in our analyses (Curtin 2002), rather than combined data for subsequent periods (Higgins 2011a). However, there were no cross‐over studies.

Studies with more than one ES group

Where multiple arms were reported in a single study, we included only the relevant arms. In studies where two or more different types of ES, or two or more types of electrode placements were compared with a control arm, we extracted data from each intervention arm but divided the control group by the number of intervention arms so that participants were not double‐counted. If the study data could not be analysed correctly, outcome data were extracted and presented but not analysed together.

Studies in which multiple observations were taken on the same individual

In studies with multiple observations for an individual, we extracted the data collected at the end of the intervention period for all analyses. For example, if ES was applied for six weeks and outcomes were measured at two, four, six, eight and 10 weeks, we used the data collected at six weeks (i.e. at the end of the intervention).

We planned to do subgroup analyses with data categorised as either (Schünemann 2011a):

short‐term effects of ES: data collected within four weeks of the completion of the intervention; or
long‐term effects of ES: data collected more than four weeks after the completion of the intervention.

However, ultimately this was not done because only one study measured outcomes more than four weeks after the completion of the intervention.

Studies in which more than one pressure ulcer was treated per participant

Where studies randomised at the participant level and measured outcomes at the pressure ulcer level (e.g. pressure ulcer healing), we treated the participant as the unit of analysis when the number of pressure ulcers assessed appeared to equal the number of participants (e.g. one pressure ulcer per participant).

We planned to incorporate cluster trials into the meta‐analyses, if the studies had been analysed correctly. Where a cluster trial was incorrectly analysed, we planned to record this as part of the 'Risk of bias' assessment. If possible, we planned to approximate the correct analyses based on the Cochrane Handbook for Systematic Reviews of Interventions guidance (Deeks 2011), using information on:

the number of clusters (or groups) randomised to each intervention group; or the average (mean) size of each cluster;
whether the outcome data ignored the cluster design for the total number of individuals (e.g. number or proportion of individuals with events, or means and SDs); and
an estimate of the intracluster (or intraclass) correlation coefficient (ICC) (Schünemann 2011a).

If studies randomised participants, but collected and reported outcome data on multiple pressure ulcers in some, but not all participants, we did not consider this a cluster trial per se, but rather a study that incorrectly included a mixture of individual and clustered data. We noted such studies and recorded the issue in the 'Risk of bias' assessment. We included these studies in the meta‐analysis but then conducted a sensitivity analysis to determine the effect of their inclusion.

Dealing with missing data

If data were not provided in numerical format and only provided in graphs, we planned to estimate the mean scores and SDs from the graphs. If studies did not provide a mean (SD) for continuous data, and it could not be derived, but studies did provide medians and interquartile ranges, we planned to extract medians and we planned to estimate the SD as 80% of the interquartile range.

If data were missing altogether, we contacted study authors. If authors did not respond or were unable to provide the additional data, we included whatever data were available. If insufficient data were available for analyses, we only presented descriptive data in the review.

If authors of trials provided both intention‐to‐treat and per protocol data, we planned to use the intention‐to‐treat data. We did not plan to impute missing data.

Assessment of heterogeneity

We considered conducting a meta‐analysis if there were at least two clinically homogenous studies (studies that investigated the effect of similar interventions on similar participant groups and reported similar outcomes). In such circumstances the I² statistic was used to quantify the statistical heterogeneity and inform decisions about whether to pool data (Higgins 2003). We considered I² values less than, or equal to 40% indicative of a low level of heterogeneity, and values that exceeded 75% indicative of a very high level of heterogeneity (Deeks 2011). We analysed data using Review Manager 5 (Review Manager 2014).

Assessment of reporting biases

We used a funnel plot to assess the possibility of small sample and reporting bias on the estimates for the effects of ES on the proportion of pressure ulcers healed (Sterne 2011).

Data synthesis

We pooled results in meta‐analyses, provided there was not excessive clinical or methodological heterogeneity, and the studies were appropriately similar in terms of type of ES, duration of pressure ulcers, type of participants, duration of treatments, and outcome assessments. Clinical and methodological heterogeneity were based on the review authors' judgement.

We presented meta‐analyses of outcome data using Review Manager 5 (Review Manager 2014). The decision to pool data in a meta‐analysis was based on the availability of outcome data and assessment of between‐trial heterogeneity. For comparisons where there was no apparent clinical heterogeneity and the I² value was less than, or equal to 40%, we pooled data using a fixed‐effect model (Demets 1987). Where there was no apparent clinical heterogeneity and the I² value was greater than 40%, we planned to pool data using a random‐effects model (DerSimonian 1986). However, we did not pool data where heterogeneity was very high (I² values of 75% or above).

We presented data using forest plots, where possible, in the following ways.

For continuous outcomes (e.g. surface area of pressure ulcers), we used the inverse variance method when summary estimates were presented as MDs with 95% CIs or SMDs with 95% CIs.
For dichotomous outcomes (e.g. proportion of pressure ulcers healed), we used the inverse variance method when summary estimates were presented as RRs with 95% CIs and we planned to use the Peto method when summary estimates were presented as odds ratios (ORs) with 95% CIs.
For time‐to‐event outcomes (e.g. time to complete healing), we used the generic inverse variance method when summary estimates were presented as HRs with 95% CIs. If HRs were not reported, but time‐to‐event data were reported, we calculated the HR with 95% CI using the methods suggested by Tierney 2007. If data were provided but could not be analysed, they were included in this review but not pooled.

For all analyses we obtained pooled estimates of treatment effect by using Review Manager 5 (Review Manager 2014).

Subgroup analysis and investigation of heterogeneity

We planned to perform subgroup analyses to explore the influence of the following variables on effect size, but these were not carried out because of an insufficient number of studies or participants.

Type of ES: (direct current versus pulsating current). We planned to explore differences in the response to the two types of ES because it is possible that different currents have different effects on healing.
Duration of pressure ulcers: (acute versus chronic i.e. less than 3 months versus more than 3 months). We planned to explore differences in the response to ES of acute and chronic pressure ulcers because acute pressure ulcers may respond better and more quickly to ES than chronic pressure ulcers.
Type of participants: (participants with spinal cord injuries versus participants without spinal cord injuries). We planned to explore differences in the response to ES of participants with spinal cord injuries versus participants without spinal cord injuries because people with spinal cord injuries have additional impairments that may influence the effectiveness of ES.
Duration of treatment effect: (short‐term treatment effect versus long‐term treatment effect i.e. effects present up to 4 weeks after the last intervention versus effects present for 4 weeks and more after the last intervention). We planned to explore differences in the duration of treatment effects because the short‐term effects of ES may differ to the long‐term effects.

Sensitivity analysis

Four sensitivity analyses were conducted to examine the robustness of the meta‐analyses to the inclusion of studies at high risk of bias from the following four domains on the 'Risk of bias' tool: the generation of the random allocation sequence, use of concealed allocation, use of blinded assessors, and dropouts (Deeks 2011). For each sensitivity analysis we excluded studies that were rated at high or unclear risk of bias. We performed additional sensitivity analyses to determine the effect of including studies with unit of analysis issues.

'Summary of findings' tables

We present the main results of the review in a 'Summary of findings’ table. This table presents key information concerning the certainty of the evidence, the magnitude of the effects of ES and the sum of available data for the main outcomes as recommended by the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2011b). The 'Summary of findings' table also includes an overall grading of the evidence related to each of the main outcomes using the GRADE approach (Schünemann 2011a). The GRADE approach defines the certainty of a body of evidence as the extent to which one can be confident that an estimate of effect is close to the true value for an outcome (Guyatt 2011; Higgins 2011a). The certainty of a body of evidence involves consideration of within‐trial risk of bias (methodological certainty), directness of evidence, heterogeneity, precision of effect estimates and risk of publication bias as recommended by the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2011a).

We present the following outcomes in the 'Summary of findings' tables.

Proportion of pressure ulcers healed.
Time to complete healing.
Complications/adverse events related to pressure ulcers.
Quality of life.

Ethics and inequalities

We addressed considerations of inequities by ensuring that we extracted data about population characteristics that are associated with health inequalities or disadvantage (Welsh 2016).

Context

We addressed contextual factors by ensuring that we extracted data about the target groups or populations (Armstrong 2011).

Results

Description of studies

See Characteristics of included studies and Characteristics of excluded studies.

Results of the search

The search generated 370 records. Forty‐five of these were duplicates, leaving 325 potentially eligible records. We retrieved 55 full‐text articles for consideration for inclusion (Figure 1) (Liberati 2009). We excluded 28 full‐text articles, two studies are awaiting classification (Feldman 2005; Karba 1995), and five studies are ongoing (ACTRN12617001534370; ACTRN12618000345280; JPRN‐UMIN000029516; NCT03753581; NTR6450). Ultimately, 20 studies met the inclusion criteria (Adegoke 2001; Adunksy 2005; Ahmad 2008; Asbjornsen 1990; Baker 1996; Carley 1985; Feeder 1991; Franek 2011; García‐Pérez 2018; Gentzkow 1991; Griffin 1991; Houghton 2010; Jercinovic 1994; Karba 1995; Kloth 1988; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018; Wood 1993). We contacted authors of 11 studies for additional information (Adegoke 2001; Adunksy 2005; Ahmad 2008; Baker 1996; Feeder 1991; Franek 2011; Houghton 2010; Karba 1995; Kloth 1988; Polak 2016a; Polak 2016b; missing data and/or any ambiguities), and we received replies from authors of seven studies (Adunksy 2005; Feeder 1991; Franek 2011; Karba 1995; Kloth 1988; Polak 2016a; Polak 2016b).

Figure 1

Study flow diagram.

Included studies

The details of the 20 included studies are provided in the Characteristics of included studies table.

Study design and setting

All included studies used a parallel‐group design. Twelve studies were single‐centred RCTs (Adegoke 2001; Asbjornsen 1990; Baker 1996; Carley 1985; Franek 2011; Griffin 1991; Houghton 2010; Jercinovic 1994; Karba 1995; Kloth 1988; Polak 2017; Polak 2018), and eight were multicentred RCTs (Adunksy 2005; Ahmad 2008; Feeder 1991; García‐Pérez 2018; Gentzkow 1991; Polak 2016a; Polak 2016b; Wood 1993). Nineteen studies were conducted in four different settings, including rehabilitation and geriatric hospitals (11 studies; Adegoke 2001; Adunksy 2005; Asbjornsen 1990; Carley 1985; Franek 2011; Gentzkow 1991; Griffin 1991; Jercinovic 1994; Karba 1995; Kloth 1988; Polak 2018), medical centres (4 studies; Baker 1996; Feeder 1991; Polak 2016a; Wood 1993), a residential care centre (2 studies; Polak 2016b; Polak 2017), and a community‐based centre (2 studies; García‐Pérez 2018; Houghton 2010). The setting of one study was unknown (Ahmad 2008). Studies were conducted in nine different countries including Canada (2 studies; Gentzkow 1991; Houghton 2010), Egypt (1 study; Ahmad 2008), Israel (1 study; Adunksy 2005), Nigeria (1 study; Adegoke 2001), Norway (1 study; Asbjornsen 1990), Poland (5 studies; Franek 2011; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018), Spain (1 study; García‐Pérez 2018), Slovenia (2 studies; Jercinovic 1994; Karba 1995), and the USA (6 studies; Baker 1996; Carley 1985; Feeder 1991; Griffin 1991; Kloth 1988; Wood 1993).

Thirteen studies did not clearly state whether participants or pressure ulcers were randomised, but the number of pressure ulcers equalled the number of participants, so there was not a unit of analysis issue (Adegoke 2001; Adunksy 2005; Ahmad 2008; Asbjornsen 1990; Carley 1985; Franek 2011; García‐Pérez 2018; Houghton 2010; Karba 1995; Kloth 1988; Polak 2016a; Polak 2017; Polak 2018). One study clearly stated that participants (not pressure ulcers) were randomised and the number of participants equalled the number of pressure ulcers, so there was not a unit of analysis issue (Griffin 1991). Four studies did not clearly state whether participants or pressure ulcers were randomised and the number of pressure ulcers was greater than the number of participants. There was no accounting for non‐independence of data in the analysis, resulting in a unit of analysis issue (Baker 1996; Feeder 1991; Jercinovic 1994; Polak 2016b). One study did not clearly state whether participants or pressure ulcers were randomised, however, the authors provided the individual participant data and from this it appeared that pressure ulcers (not participants) were randomised. There was no accounting for non‐independence of data in the analysis, resulting in a unit of analysis issue (Wood 1993). One study clearly stated that pressure ulcers (not participants) were randomised. There was no accounting for non‐independence of data in the analysis, resulting in a unit of analysis issue (Gentzkow 1991).

Participants

A total of 913 participants were randomised with sample sizes ranging from seven participants in Adegoke 2001 to 80 participants in Baker 1996. The mean age of the participants in the included studies ranged from 26 years to 83 years. Overall, 50% of participants were male. The chronicity of the pressure ulcers was variable, ranging from a mean of 4 days in Adunksy 2005 to more than 12 months in Feeder 1991. In 16 studies, pressure ulcers were on the sacral and coccygeal region (30%), ischium (24%), lower extremities including heels (23%), greater trochanter of the femur (7%), and other parts of the body (4%). Four studies did not provide information about the location of the pressure ulcers (Ahmad 2008; Carley 1985; Karba 1995; Kloth 1988). Fourteen studies provided data on the severity of the pressure ulcers. In these studies, most participants had stage II (37%) or stage III (45%) pressure ulcers. Six studies did not provide information about the severity of the pressure ulcers (Asbjornsen 1990; Baker 1996; Carley 1985; Jercinovic 1994; Karba 1995; Wood 1993). Two studies had participants with wounds from different causes (Baker 1996; Feeder 1991). In both studies, more than 75% of the wounds were pressure ulcers, but neither study provided individual participant data.

Interventions

Electrical stimulation (ES) was administered from two to 20 hours per week (median 5, interquartile range 4 to 8) and for between three and 12 weeks (median 6, interquartile range 4 to 8). Four studies administered direct current (Adegoke 2001; Adunksy 2005; Carley 1985; Griffin 1991), and 16 studies administered pulsating current (Ahmad 2008; Asbjornsen 1990; Baker 1996; Feeder 1991; Franek 2011; García‐Pérez 2018; Gentzkow 1991; Houghton 2010; Jercinovic 1994; Karba 1995; Kloth 1988; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018; Wood 1993).

All studies, but one, used two electrodes for the administration of ES (i.e. an active electrode and a dispersive electrode). Electrodes were placed in three different ways, namely:

in 13 studies, one electrode was placed over the treating pressure ulcer and the other electrode was placed on healthy skin next to the pressure ulcer (Adegoke 2001; Ahmad 2008; Carley 1985; Feeder 1991; Franek 2011; Gentzkow 1991; Griffin 1991; Houghton 2010; Kloth 1988; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018);
in five studies, both electrodes were placed on healthy skin around the pressure ulcer (Adunksy 2005; Baker 1996; Jercinovic 1994; Karba 1995; Wood 1993);
in one study, the two electrodes were placed on either side of one hand for pressure ulcers on the sacrum and heel (Asbjornsen 1990); and
in one study, the four electrodes were placed around the ulcer (García‐Pérez 2018).

Studies used different current intensities, namely:

in nine studies, the intensity was set to elicit a visible minimal motor contraction (Adegoke 2001; Ahmad 2008; Asbjornsen 1990; Baker 1996; Griffin 1991; Houghton 2010; Jercinovic 1994; Karba 1995; Kloth 1988); and
in six studies, the intensity was set to elicit a mild tingling sensation (Franek 2011; García‐Pérez 2018; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018).

Studies used different frequencies, namely:

in five studies, frequency was set to less than or equal to 50 Hz (Adegoke 2001; Baker 1996; García‐Pérez 2018; Jercinovic 1994; Wood 1993);
in seven studies, frequency was set between 50 Hz and 100 Hz, inclusive (Asbjornsen 1990; Franek 2011; Griffin 1991; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018);
in two studies, frequency was set to more than 100 Hz (Ahmad 2008; Kloth 1988);
in three studies, frequency was changed over the course of the study (Feeder 1991; Gentzkow 1991; Houghton 2010), and one study did not report the frequency (Karba 1995); and
two studies used direct current, therefore frequency was not applicable (Adunksy 2005; Carley 1985).

Five studies did not provide information about the intensity of the current (Adunksy 2005; Carley 1985; Feeder 1991; Gentzkow 1991; Wood 1993).

Fourteen studies used a placebo or sham ES as the control (Adegoke 2001; Adunksy 2005; Ahmad 2008; Asbjornsen 1990; Baker 1996; Feeder 1991; Gentzkow 1991; Griffin 1991; Karba 1995; Kloth 1988; Polak 2016a; Polak 2017; Polak 2018; Wood 1993), and six studies had no type of ES as the control (Carley 1985; Franek 2011; García‐Pérez 2018; Houghton 2010; Jercinovic 1994; Polak 2016b). Standard nursing or wound care was provided to all groups in all studies.

Outcomes

Seventeen studies (out of 20) provided data on five of our outcomes of interest. Twelve studies measured the proportion of pressure ulcers healed, one measured pressure ulcer severity on a composite measure, 13 measured the surface area of pressure ulcers, five measured the time to complete healing, 14 measured the rate of pressure ulcer healing, and 13 reported adverse events. Ten studies did not provide useable data for one or more of our outcomes of interest (Adegoke 2001; Adunksy 2005; Asbjornsen 1990; Baker 1996; Feeder 1991; Gentzkow 1991; Griffin 1991; Houghton 2010; Kloth 1988; Polak 2017).

Registry and funding source

Two studies were prospectively registered (Polak 2017; Polak 2018), and two studies were retrospectively registered (Polak 2016a; Polak 2016b), on the Australian and New Zealand Clinical Trials Registry. Eight studies received full or partial funding from medical device companies (Adunksy 2005; Carley 1985; Gentzkow 1991; Griffin 1991; Houghton 2010; Karba 1995; Kloth 1988; Wood 1993), four studies received full or partial funding from their institutions (Polak 2016b; Polak 2017; Polak 2018; Wood 1993), four studies received funding from research grants (Baker 1996; Houghton 2010; Jercinovic 1994; Karba 1995), three studies did not receive any funding (Franek 2011; García‐Pérez 2018; Polak 2016a), and four studies did not provide information about any source of funding (Adegoke 2001; Ahmad 2008; Asbjornsen 1990; Feeder 1991).

Excluded studies

We excluded a total of 28 full‐text articles for one or more of the following reasons (see Characteristics of excluded studies).

Study design (20 studies): these studies were excluded because they were before‐and‐after intervention studies, prospective non‐randomised intervention studies, reviews, economic analyses (Allen 2004; Barczak 2001; Barron 1985; Chalker 1983; Clegg 2007; Cukjati 2001; Edsberg 2002; Gault 1976; Gentzkow 1993; Karsli 2017; Koel 2014; Lawson 2007; Lee 2007; Lippert‐Gruner 2003; Polak 2014; Recio 2012; Stefanovska 1993; Trontelj 1994; Ullah 2007; Wolcott 1969).
Population (6 studies): these studies were excluded because they were either preclinical or included participants with leg ulcers or other types of wounds (Goldman 2004; Houghton 2003; Jankovic 2008; Sugimoto 2012; Van Londen 2008; Yoshikawa 2015).
Intervention (2 studies): these studies were excluded because they assessed electromagnetic therapy or acupuncture for the treatment of pressure ulcers (Comorosan 1993; Jia 2015).

Studies awaiting classification

There are two studies awaiting classification (Feldman 2005; Karba 1997; see Characteristics of studies awaiting classification).

Ongoing studies

There are five ongoing studies (ACTRN12617001534370; ACTRN12618000345280; JPRN‐UMIN000029516; NCT03753581; NTR6450; see Characteristics of ongoing studies).

Risk of bias in included studies

We assessed all 20 included studies for risk of bias across the eight domains. The results are shown in Figure 2 and Figure 3 with judgements explained in the Characteristics of included studies table.

Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Random sequence generation

Eight studies had a low risk of bias for this domain. These studies used computer software or coin tossing to generate their randomisation sequences (Adunksy 2005; Franek 2011; García‐Pérez 2018; Houghton 2010; Kloth 1988; Polak 2016b; Polak 2017; Polak 2018).
Twelve studies had an unclear risk of bias for this domain. These studies did not provide sufficient details to make a judgement (Adegoke 2001; Ahmad 2008; Asbjornsen 1990; Baker 1996; Carley 1985; Feeder 1991; Gentzkow 1991; Griffin 1991; Jercinovic 1994; Karba 1995; Polak 2016a; Wood 1993).
No studies had a high risk of bias for this domain.

Concealed allocation

Nine studies had a low risk of bias for this domain. These studies reported that participants' allocation to groups was either done by an independent person or with the use of opaque sealed sequentially numbered envelopes (Adegoke 2001; Adunksy 2005; Franek 2011; Houghton 2010; Kloth 1988; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018).
Ten studies had an unclear risk of bias for this domain. These studies did not provide sufficient details to make a judgement (Ahmad 2008; Asbjornsen 1990; Baker 1996; Carley 1985; Feeder 1991; Gentzkow 1991; Griffin 1991; Jercinovic 1994; Karba 1995; Wood 1993).
One study had a high risk of bias for this domain. This study stated that the study personnel who determined eligibility also prepared the randomisation sequence (García‐Pérez 2018).

Blinding

Blinding of participants

Fourteen studies had a low risk of bias for this domain. These studies provided control participants with a placebo or sham ES treatment (Adegoke 2001; Adunksy 2005; Ahmad 2008; Asbjornsen 1990; Baker 1996; Feeder 1991; Gentzkow 1991; Griffin 1991; Karba 1995; Kloth 1988; Polak 2016a; Polak 2017; Polak 2018; Wood 1993).
No studies had an unclear risk of bias for this domain.
Six studies had a high risk of bias for this domain. These studies could not blind participants because there was no placebo or sham ES treatment for control participants (Carley 1985; Franek 2011; García‐Pérez 2018; Houghton 2010; Jercinovic 1994; Polak 2016b).

Blinding of personnel

Five studies had a low risk of bias for this domain. These studies clearly reported that personnel were blinded to the treatment group or were unaware of the participants’ allocation (Adunksy 2005; Feeder 1991; Gentzkow 1991; Kloth 1988; Wood 1993).
Three studies had an unclear risk of bias for this domain. These studies did not provide sufficient details to make a judgement (Ahmad 2008; Asbjornsen 1990; Karba 1995).
Twelve studies had a high risk of bias for this domain. These studies could not blind personnel because control participants did not receive placebo or sham ES. Alternatively, these studies clearly stated that it was not possible to blind personnel even though control participants received placebo or sham ES treatment (Adegoke 2001; Baker 1996; Carley 1985; Franek 2011; García‐Pérez 2018; Griffin 1991; Houghton 2010; Jercinovic 1994; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018).

Blinding of outcome assessment

Eleven studies had a low risk of bias for this domain. These studies clearly reported blinding of outcome assessors to group allocation (Adunksy 2005; Asbjornsen 1990; Feeder 1991; Franek 2011; García‐Pérez 2018; Houghton 2010; Kloth 1988; Polak 2016a; Polak 2017; Polak 2018; Wood 1993).
Seven studies had an unclear risk of bias for this domain. These studies did not provide sufficient details to make a judgement (Adegoke 2001; Ahmad 2008; Carley 1985; Gentzkow 1991; Jercinovic 1994; Karba 1995; Polak 2016b).
Two studies had a high risk of bias for this domain. These studies clearly reported that the outcome assessors were not blinded (Baker 1996; Griffin 1991).

Incomplete outcome data

Thirteen studies had a low risk of bias for this domain. These studies had a dropout rate of less than 15% (Adegoke 2001; Ahmad 2008; Carley 1985; Franek 2011; García‐Pérez 2018; Griffin 1991; Houghton 2010; Jercinovic 1994; Karba 1995; Kloth 1988; Polak 2016b: Polak 2017; Wood 1993).
One study had an unclear risk of bias for this domain. This study did not provide sufficient details to make a judgement (Baker 1996).
Six studies had a high risk of bias for this domain. These studies had a dropout rate of between 18% and 40% (Adunksy 2005; Asbjornsen 1990; Feeder 1991; Gentzkow 1991; Polak 2016a; Polak 2018).

Selective reporting

Sixteen studies had a low risk of bias for this domain. These studies reported data on all outcomes stated in the methods (Adegoke 2001; Adunksy 2005; Ahmad 2008; Asbjornsen 1990; Carley 1985; Franek 2011; García‐Pérez 2018; Gentzkow 1991; Griffin 1991; Karba 1995; Kloth 1988; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018; Wood 1993).
No studies had an unclear risk of bias for this domain.
Four studies had a high risk of bias for this domain. These studies either reported outcome data for only a specific group of those randomised or did not report data on all outcomes stated in the methods (Baker 1996; Feeder 1991; Houghton 2010; Jercinovic 1994).

Other potential sources of bias

Seven studies had a low risk of bias for this domain. These studies were free of other sources of potential bias (Adegoke 2001; Ahmad 2008; Asbjornsen 1990; García‐Pérez 2018; Polak 2016a; Polak 2017; Polak 2018).
Three studies had an unclear risk of bias for this domain. These studies did not provide sufficient details to make a judgement but there were sufficient reasons to believe that there may be other sources of bias (Carley 1985; Franek 2011; Griffin 1991).
Ten studies had a high risk of bias for this domain. These studies had potential bias due to some aspect of study design. This included extreme baseline imbalance or unit of analysis issues (i.e. participants with multiple pressure ulcers were recruited and data were presented at the pressure ulcer level rather than participant level) (Adunksy 2005; Baker 1996; Feeder 1991; Gentzkow 1991; Houghton 2010; Jercinovic 1994; Karba 1995; Kloth 1988; Polak 2016b; Wood 1993). Alternatively (or in addition), these studies were sponsored by industry (Adunksy 2005; Carley 1985; Gentzkow 1991; Griffin 1991; Houghton 2010; Kloth 1988; Wood 1993).

Effects of interventions

See: Summary of findings for the main comparison Electrical stimulation (plus standard care) versus sham/no ES (plus standard care) for treating pressure ulcers

All included studies compared a type of ES (plus standard care) with sham, placebo or no ES (plus standard care). We have not attempted to distinguish between placebo and sham because the two terms are used interchangeably by authors of included studies. Standard care included any of the following: wound dressings, pressure relief, regular turning, nutritional advice, and nutritional supplements. The studies administered standard care in the same manner to both groups.

All studies included a measure of at least one of the outcomes of interest. They examined the proportion of pressure ulcers healed, pressure ulcer severity on a composite measure, surface area of pressure ulcers, time to complete healing, complication/adverse events and rate of pressure ulcer healing. One study included a composite measure of pressure ulcer severity but did not provide useable data. No study included measures reflective of three of the outcomes of interest, including quality of life, depression and consumers' perceptions of treatment effectiveness. The results of all analyses are reported below.

Electrical stimulation (plus standard care) versus sham/no ES (plus standard care)

Primary outcome: proportion of pressure ulcers healed

Twelve studies with a total of 581 participants (697 pressure ulcers) examined the proportion of pressure ulcers healed (Adunksy 2005; Asbjornsen 1990; Baker 1996; Feeder 1991; Franek 2011; Griffin 1991; Houghton 2010; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018; Wood 1993). The data in all these studies were expressed as the number of pressure ulcers healed. Eleven studies with a total of 501 participants (512 pressure ulcers) provided sufficient data for meta‐analysis (Adunksy 2005; Asbjornsen 1990; Feeder 1991; Franek 2011; Griffin 1991; Houghton 2010; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018; Wood 1993), and were pooled using a fixed‐effect model. ES probably increases the proportion of pressure ulcers healed when compared with no ES (risk ratio (RR) 1.99, 95% confidence interval (CI) 1.39 to 2.85; I² = 0%; Analysis 1.1; summary of findings Table for the main comparison). We downgraded the evidence to moderate certainty for serious risk of bias (because a lot of the studies had either high or unclear risk of bias for performance bias and selective reporting). Importantly, three of the included studies did not account for non‐independence of data, resulting in unit of analyses issues (Feeder 1991; Polak 2016b; Wood 1993). Two of these studies randomised at the participant level, but included a few participants with more than one pressure ulcer and analysed data at the pressure ulcer, not participant level, without taking into account the clustered nature of data (Feeder 1991; Polak 2016b). One of these studies randomised at the pressure ulcer level and included a few participants with more than one pressure ulcer, and analysed data at the pressure ulcer level without taking into account the non‐independence of data (Wood 1993). We performed a sensitivity analysis to determine the effect of these three studies on the overall estimate; they made little difference in the overall treatment effect (RR 1.99, 95% CI 1.39 to 2.85 versus RR 1.79, 95% CI 1.17 to 2.73) although as expected the estimate is less precise with the removal of the three studies.

Primary outcome: composite measures of pressure ulcer severity

Two studies with a combined total of 51 participants (51 pressure ulcers) reported a composite measure of pressure ulcer severity (García‐Pérez 2018; Houghton 2010). The data in both these studies were expressed as a composite number. One study used the photographic wound assessment tool (Houghton 2010), but did not provide sufficient data to be included in the analyses. The other study used the Resultados Esperados de la Valoracion y Evolucion de la Cicatrizacion de las Heridas cronicas (RESVECH) Index (García‐Pérez 2018). The point estimate for the mean difference (MD) was ‐2.43 points (95% CI ‐6.14 to 1.28; Analysis 1.2).

Primary outcome: surface area of pressure ulcers

Fourteen studies with a total of 590 participants (706 pressure ulcers) examined the surface area of pressure ulcers (Adegoke 2001; Adunksy 2005; Ahmad 2008; Asbjornsen 1990; Baker 1996; Feeder 1991; Franek 2011; García‐Pérez 2018; Karba 1995; Kloth 1988; Polak 2016b; Polak 2017; Polak 2018; Wood 1993). The data in all these studies were expressed as mm² or cm². For the purpose of analyses we converted surface area into cm². Twelve studies with a total of 494 participants (505 pressure ulcers) provided sufficient data (Adegoke 2001; Adunksy 2005; Ahmad 2008; Asbjornsen 1990; Feeder 1991; Franek 2011; García‐Pérez 2018; Karba 1995; Polak 2016b; Polak 2017; Polak 2018; Wood 1993). We did not pool the data because there was considerable statistical heterogeneity between studies (I² = 96%). The source of the heterogeneity was not apparent, but is probably due to a variety of factors, including differences in the types of participants, length of intervention, duration of pressure ulcers and different risks of bias. It is uncertain whether ES decreases the surface area of pressure ulcers when compared with no ES. The MD for each included study is presented as Analysis 1.3. The point estimates for the MD of each study ranged from ‐0.90 cm² to 10.37 cm². We did not include this outcome in the 'Summary of findings' table but nonetheless rated it using the GRADE assessment of the certainty of evidence. We downgraded the evidence to very low certainty: downgrading once for serious risk of bias (because a lot of the studies had either high or unclear risk of bias for selection and detection bias), once for inconsistency and once for imprecision.

Primary outcome: time to complete healing

Five studies with a total of 181 participants (184 pressure ulcers) examined time to complete healing (Adunksy 2005; Asbjornsen 1990; Feeder 1991; Griffin 1991; Polak 2017). The data in these studies were expressed as number of days to complete healing. Two studies with a total of 55 participants (55 pressure ulcers) provided sufficient data (Adunksy 2005; Griffin 1991), and were pooled using a fixed‐effect model. It is uncertain whether ES decreases the time to complete healing of pressure ulcers compared with no ES (hazard ratio (HR) 1.06, 95% CI 0.47 to 2.41; I² = 0%; Analysis 1.4). We downgraded the evidence to very low certainty: once for serious risk of bias (because both studies had high risk of bias for 2 domains and 1 study had unclear risk of bias for another 3 domains), once for indirectness (because the 2 studies were not reflective of all who are vulnerable to pressure ulcers) and twice for imprecision (summary of findings Table for the main comparison).

Primary outcome: complications/adverse events

Thirteen studies with a total of 586 participants (602 pressure ulcers) provided statements about adverse events (Adunksy 2005; Asbjornsen 1990; Carley 1985; Feeder 1991; Franek 2011; García‐Pérez 2018; Gentzkow 1991; Griffin 1991; Houghton 2010; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018). However, the data were not sufficiently detailed or comparable to analyse quantitatively. We downgraded the evidence to low certainty: once for serious risk of bias (because a lot of the studies had either high or unclear risk of bias for selection and attrition bias) and once for imprecision (summary of findings Table for the main comparison). The narrative descriptions of the adverse events in these 13 studies are therefore provided below.

Adunksy 2005 reported that 25 participants withdrew. Ten participants (5 in the experimental group and 5 in the control group) withdrew for a variety of medical reasons. Another 15 participants withdrew because of other adverse events, such as limb amputation (3 participants), deterioration of the pressure ulcer (1 participant), other medical problems (pneumonia, urosepsis, ischaemic colitis, installation of a cardiac pacemaker; 8 participants), or other reasons (3 participants). In addition to the withdrawals, there were four adverse events in two participants from the experimental group (excessive granulation and local irritation to the ES).
Asbjornsen 1990 reported two adverse events (1 leg amputation in the experimental group, and 1 death in the control group).
Carley 1985 reported no complications/adverse events for participants in the experimental group. They however stated that "the control wounds would typically redevelop eschars that required repeated debridement as often as every two weeks" (p444). This was associated with pain and discomfort.
Feeder 1991 stated that 15% of participants had minor uncomfortable tingling in the wound bed (20% of participants in the experimental group and 10% of participants in the control (sham ES) group).
Franek 2011 stated that three participants had complications that were not related to the intervention, including one death (2 in the experimental group and 1 in the control group).
Gentzkow 1991 stated that 14% of participants in the experimental group and 4% of participants in the control group had occasional uncomfortable sensations in the wound bed.
Griffin 1991 stated that three participants (2 in the experimental group and 1 in the control group) withdrew from the study because of medical complications (n = 2) and need for surgical repair of the pressure ulcer (n = 1).
Houghton 2010 stated that the adverse events were minor for participants in the experiment group. For example, two participants in the experiment group had red, raised and itchy skin under one of the electrodes (lasting more than 24 hours and less than 48 hours). A third participant in the experimental group complained of dizziness and delusions. These were not attributed to the intervention. No information on adverse events was provided for participants in the control group.
García‐Pérez 2018, Polak 2016a, Polak 2016b, Polak 2017 and Polak 2018 stated that no adverse events were observed in the experimental group/s.

Secondary outcome: rate of pressure ulcer healing

Fourteen studies with a total of 657 participants (812 pressure ulcers) examined the rate of pressure ulcer healing (Baker 1996; Feeder 1991; Franek 2011; Gentzkow 1991; Griffin 1991; Houghton 2010; Jercinovic 1994; Karba 1995; Kloth 1988; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018; Wood 1993). The data in these studies were expressed in various ways including percentage healed per day, percentage healed per week, percentage healed per four weeks, percentage healed per six weeks and percentage healed per eight weeks. For the purpose of analysis, we expressed data as percentage healed per week by dividing or multiplying the data by an appropriate conversion constant (e.g. percentage healed per day was multiplied by 7 and percentage healed per 4 weeks was divided by 4 to obtain percentage healed per week). Twelve studies with a total of 561 participants (613 pressure ulcers) provided sufficient data (Feeder 1991; Franek 2011; Gentzkow 1991; Griffin 1991; Houghton 2010; Jercinovic 1994; Karba 1995; Polak 2016a; Polak 2016b; Polak 2017; Polak 2018; Wood 1993), and we analysed these studies using a fixed‐effect model. ES probably increases the rate of pressure ulcer healing when compared with no ES (MD 4.59% per week, 95% CI 3.49% to 5.69%; I² = 25%; Analysis 1.5). This outcome was not included in the 'Summary of findings' table but we nonetheless rated this using the GRADE assessment of the certainty of evidence. We downgraded the evidence to moderate certainty: twice for serious risk of bias (because a lot of the studies had either high or unclear risk of bias for selection bias and some studies had high risk of bias for attrition and reporting bias). Importantly, five of the included studies did not account for non‐independence of data, resulting in unit of analyses issues (Feeder 1991; Gentzkow 1991; Jercinovic 1994; Polak 2016b; Wood 1993). Three of these studies randomised at the participant level, but included a few participants with more than one pressure ulcer and analysed data at the pressure ulcer, not participant level, without taking into account the clustered nature of the data (Feeder 1991; Jercinovic 1994; Polak 2016b). Two of these studies randomised at the pressure ulcer level and included a few participants with more than one pressure ulcer, and analysed data at the pressure ulcer level, without taking into account the non‐independence of data (Gentzkow 1991; Wood 1993). We performed a sensitivity analysis to determine the effect of including these five studies. The results indicated very little difference in treatment effect with or without these five studies (MD 4.59% per week, 95% CI 3.49% to 5.69% versus MD 4.21% per week, 95% CI 3.03% to 5.40%).

Secondary outcome: quality of life, depression and consumers' perceptions of treatment effectiveness

No studies were found that measured quality of life, depression or consumers' perceptions of treatment effectiveness.

Subgroup analyses

We did not perform subgroup analyses to explore the influence of the variables on effect size because of an insufficient number of studies or participants.

Sensitivity analysis

We conducted sensitivity analyses to examine the effects of the randomisation process (adequate sequence generation versus inadequate sequence generation), concealed allocation (concealed allocation versus non‐concealed allocation), blinding of assessors (blinding of assessors versus no blinding of assessors) and dropouts (more than 15% dropouts versus 15% or less dropouts) on the primary outcome of proportion of pressure ulcers healed. For each analysis, we excluded between three and five studies (out of 12 studies) because of high or unclear risk. These exclusions had no effect on the RR (see Table 1). We performed additional sensitivity analyses to determine the effect of including studies with unit of analyses issues (these results are reported for each relevant outcome in the results section above).

Open in table viewer

Table 1. Sensitivity analyses‐ bias

Outcomes

Pooled results

Randomisation (studies with
adequate sequence generation)

Allocation (studies with
concealed allocation)

Assessors (studies with
blinded assessors)

Dropout rate (studies
with < 15% dropouts)

Proportion of

pressure ulcers healed

1.99 (1.39 to 2.85)

(n = 12)

2.12 (1.36 to 3.30)

(n = 6)

1.98 (1.35 to 2.90)

(n = 7)

1.93 (1.26 to 2.95)

(n = 9)

2.34 (1.47 to 3.71)

(n = 6)

Results are presented as RR (95% CI)
n = number of studies included in analysis

Small sample bias

We assessed the possibility of small sample and reporting bias on the estimates for the effects of ES on the proportion of pressure ulcers healed using a funnel plot (see Figure 4). There is no indication of small sample or reporting bias (Sterne 2011).

Figure 4

Funnel plot of comparison: 1 Electrical stimulation (plus standard care) versus sham/no electrical stimulation (plus standard care), outcome: 1.1 Proportion of pressure ulcers healed.

Discussion

available in

Summary of main results

summary of findings Table for the main comparison

We identified 20 eligible studies. There is moderate certainty evidence that electrical stimulation (ES) probably increases the proportion of pressure ulcers healed compared with no ES.

Two studies examined a composite measure of pressure ulcer severity. One used a photographic wound assessment tool, but the data were not useable (Houghton 2010), and it was not possible to determine the effect of ES on this outcome.

There is uncertainty as to whether ES reduces the surface area of pressure ulcers. We could not pool the data for this outcome because there was considerable statistical heterogeneity between studies (I² = 96%), and the certainty of the evidence for this outcome is very low.

Similarly, there is uncertainty as to whether ES decreases the time to complete healing of pressure ulcers when compared with no ES. The 95% confidence interval associated with the hazard ratio extends from 0.47 (favouring the control) to 2.41 (favouring ES), reflecting the uncertainty; the certainty of evidence for this outcome is very low.

Adverse events were poorly reported and included redness of the skin, itchy skin, dizziness and delusions, deterioration of the pressure ulcer, limb amputation, and occasionally death; the certainty of evidence for this outcome is low and it is difficult to attribute adverse events to interventions.

ES probably increases the rate of pressure ulcer healing when compared with no ES (moderate certainty evidence). We did not include this outcome in the 'Summary of findings' table, but we nonetheless rated it using the GRADE assessment of the certainty of evidence.

No studies included measures of quality of life, depression, or consumers' perceptions of treatment effectiveness.

Overall completeness and applicability of evidence

In some respects this review is reasonably complete and the findings are generalisable. For example, it included participants reflective of our target population. That is, participants were of both genders, various ages, and had a variety of conditions (e.g. spinal cord injury, frail older people) who were recruited from in‐ and outpatient settings. Most of the studies were recent and from different countries. In addition, the ES was applied in a way that is reflective of current practice. For example, the electrodes were applied over or around the pressure ulcers. There was one exception where the electrodes were placed on either side of one hand for pressure ulcers on the sacrum and heel (Asbjornsen 1990). However, this study only contributed to two meta‐analyses and its point estimates were similar to the other studies, suggesting that electrode placement may make little difference. This would seem surprising because presumably the mechanism of action for ES varies depending on electrode placement. More studies are required to explore this further.

However, in other aspects there are some limitations to the completeness of this review. For example, the included studies either did not provide useable data or did not assess some of the outcomes of interest to this review, such as composite measures for pressure ulcer severity, quality of life, depression, and consumers' perspective of treatment effectiveness. In addition, most studies included in this review only investigated the effects of ES over a relatively short period of time (between 20 days to 8 weeks). These aspects limit the generalisability of the results. The completeness of our review is also limited because we did not include children. Future reviews could consider expanding the inclusion criteria to children with pressure ulcers.

Certainty of the evidence

The certainty of evidence for the two healing outcomes rated in the 'Summary of findings' table was either moderate (proportion of pressure ulcers healed) or very low (time to complete healing). We downgraded the evidence mostly because of the serious risk of publication bias and because the included studies were at high or unclear risk of bias (Figure 2; Figure 3). We downgraded one outcome (surface area of pressure ulcers) for imprecision, even though we did not pool the results because the point estimates of the included studies were imprecise. Some of the more serious risks of bias in the included studies were failure to blind therapists (60% of studies), incomplete outcome data (30% of studies) and failure to blind assessors to outcomes (10% of studies). There were also other potential risks of bias (50% of studies). However, we included results from all studies in the main analyses regardless of their risk of bias.

The data extraction for this systematic review was challenging. This was due to poor reporting and the lack of consistency between authors in the way outcome data were expressed. This was particularly problematic for the rate of healing data with some authors expressing this as percentage change while others expressed it as percentage healed. For example, a change in a pressure ulcer size from 10 cm² to 8 cm² over one month was expressed by some as a reduction to 80% of original pressure ulcer size and by others as a 20% improvement. It was often difficult to ascertain how the data were expressed. In addition, the units varied between studies with some reporting rate of healing per day and others reporting rate of healing per week or per month. We dealt with this later issue by converting all different units into percentage rate of healing per week, where for example, a rate of healing of 5% reflects a 5% reduction each week with respect to the size of the pressure ulcer at the beginning of the week. In all, we would encourage readers to interpret the results of this outcome cautiously. We would strongly recommend better reporting of this outcome in future studies.

Six studies did not account for non‐independence of data, resulting in unit of analyses issues (Baker 1996; Feeder 1991; Gentzkow 1991; Jercinovic 1994; Polak 2016b; Wood 1993). Three of these studies randomised at the participant level, but included a few participants with more than one pressure ulcer and analysed data at the pressure ulcer, not participant level, without taking into account the clustered nature of the data (Feeder 1991; Jercinovic 1994; Polak 2016b). Two of these studies randomised at the pressure ulcer level and included a few participants with more than one pressure ulcer, and analysed data at the pressure ulcer level without taking into account the non‐independence of data (Gentzkow 1991; Wood 1993). We rated all of these studies as high risk on the 'other bias' domain in the 'Risk of bias' tool. Importantly, we included five of these six studies in one of the meta‐analyses, i.e. rate of pressure ulcer healing (Feeder 1991; Gentzkow 1991; Jercinovic 1994; Polak 2016b; Wood 1993), and three of these six studies were included in another meta‐analysis, i.e. proportion of pressure ulcer healed (Feeder 1991; Polak 2016b; Wood 1993). In these two meta‐analyses, the number of pressure ulcers, not the number of participants, was used to estimate the pooled effect without adjustment for the non‐independence of the data. This was done because the difference between the number of participants and number of pressure ulcers was only small and therefore unlikely to influence the results. The potential for bias in these two analyses due to this problem was captured on the GRADE risk of bias outcome. In addition, our sensitivity analyses to determine the influence of the addition of these studies on the estimates indicated that they had little effect.

Potential biases in the review process

There were two main sources of potential bias in our review process.

We attempted to contact authors for missing details but not all responded. Our failure to gather all missing data may have introduced bias.
We may have missed some potentially eligible studies. We were most likely to have missed unpublished studies, studies in languages other than English, and studies with negative results. However, there was no evidence of publication bias from the funnel plot.

Agreements and disagreements with other studies or reviews

There are five notable systematic reviews which have examined the effect of ES for the treatment of pressure ulcers (Cullum 2001; Koel 2014; Lala 2016; Reddy 2008; Vélez‐Díaz‐Pallarés 2015). We broadly agree with the interpretation of three of these reviews (Cullum 2001; Lala 2016; Vélez‐Díaz‐Pallarés 2015), despite discrepancies in data extraction between two of these reviews and our own (Lala 2016; Vélez‐Díaz‐Pallarés 2015). They, like us, conclude that there is uncertainty as to whether ES is effective. We however, do not agree with the findings of two of the five reviews (Koel 2014; Reddy 2008);

In Koel 2014, the authors examined the rate of healing, but included any type of wound, not just pressure ulcers. The pooled estimate is surprisingly similar to our pooled estimate given there are some notable discrepancies between the data extraction of the studies in common. Nonetheless, the authors made a definitive conclusion stating that: "a clear and positive recommendation is available regarding the effectiveness of ES to increase wound healing" (Koel 2014, pg. 124). We do not agree that the evidence justifies such a strong and clear conclusion.

In the second review that we disagree with (Reddy 2008), the authors examined the rate of healing of pressure ulcers. The authors of this review did not pool data because of high clinical heterogeneity. Interestingly, they concluded that: "Among the good quality RCTs examining adjunctive therapies [electric current], there were no benefits to the interventions, which included electric current (vs placebo electric current)…..." (Reddy 2008, pg. 2658). Their conclusions are in stark contrast to the conclusions of the other reviews in this area and not fully supported by our findings. Our findings indicate uncertainty, but do not indicate that ES is ineffective.

There are also some reviews that have not extracted between‐group differences and are therefore difficult to summarise (Gardner 1999; Houghton 2014; Kawasaki 2014; Lampe 1998; Polak 2014; Regan 2009; Thakral 2013). Regardless, they differ to our review by concluding that ES is effective.

In addition to systematic reviews, there are six key guidelines (AWMA 2012; Consortium for Spinal Cord Medicine 2014; Houghton 2013; NICE 2014; NPUAP/EPUAP 2014; SCIRE 2014). We agree with the recommendations from the National Institute for Health and Care Excellence (NICE 2014). They state: "Do not offer… [electrical stimulation] to adults to treat a pressure ulcer". However, they acknowledge, like we do, that this recommendation is based on very low or low GRADE evidence.

Surprisingly, most of the other guidelines state the opposite and recommend the use of ES, and all grade the certainty of evidence as high. For example, The Pan Pacific Clinical Practice Guideline for the Prevention and Management of Pressure Injury (AWMA 2012) state: "Consider using electrotherapy as an adjunct for promoting healing in pressure injuries". They rate this recommendation as level "B" according to the National Health and Medical Research Council (NHMRC) grading matrix (good evidence ‐ body of evidence can be trusted to guide practice in most situations).

In even greater contrast to our findings, The Canadian Best Practice Guidelines for the Prevention and Management of Pressure Ulcers in People With Spinal Cord Injury (Houghton 2013), emphatically state: "Use electrical stimulation combined with standard wound care interventions to promote closure of stage III or IV pressure ulcers". They rate this recommendation as "1A” according to the grading of the Registered Nurses' Association of Ontario level of evidence (evidence from meta‐analysis or systematic review of randomised controlled trials). We do not believe this rating or recommendation is justified.

The Consortium for Spinal Cord Medicine 2014 provide a similar clear directive to use ES, stating: "Use electrical stimulation (ES) to promote closure of category/stage III or IV pressure ulcers, unless contraindicated in the cases of untreated, underlying osteomyelitis or infection". They state that this recommendation is based on grade "A" according to the grading of Sackett 1989 (supported by direct scientific evidence from properly designed and implemented controlled trials, i.e. large randomised trial(s) with clear‐cut results and low risk of error).

The Spinal Cord Injury Research Evidence similarly states (SCIRE 2014): "Electrical stimulation added to standard wound management promotes healing of Stage III and IV pressure ulcers post‐SCI [spinal cord injury]". They also base their recommendation on level "1" evidence according to the grading of Sackett 1989 (supported by direct scientific evidence from properly designed and implemented controlled trials, i.e. large randomised trial(s) with clear‐cut results and low risk of error).

The results of our systematic review do not support any of these interpretations of the evidence. We conclude that the certainty of the evidence is moderate, low or very low, and not sufficient to support promoting the use of ES in the clinical setting outside of further research.

Figure 1

Study flow diagram.

Navigate to figure in ReviewOpen in new tab

Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Navigate to figure in ReviewOpen in new tab

Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Navigate to figure in ReviewOpen in new tab

Figure 4

Funnel plot of comparison: 1 Electrical stimulation (plus standard care) versus sham/no electrical stimulation (plus standard care), outcome: 1.1 Proportion of pressure ulcers healed.

Navigate to figure in ReviewOpen in new tab

Analysis 1.1

Comparison 1 Electrical stimulation (plus standard care) versus sham/no electrical stimulation (plus standard care), Outcome 1 Proportion of pressure ulcers healed.

Navigate to figure in ReviewOpen in new tab

Analysis 1.2

Comparison 1 Electrical stimulation (plus standard care) versus sham/no electrical stimulation (plus standard care), Outcome 2 Composite measure of pressure ulcer severity.

Navigate to figure in ReviewOpen in new tab

Analysis 1.3

Comparison 1 Electrical stimulation (plus standard care) versus sham/no electrical stimulation (plus standard care), Outcome 3 Surface area of pressure ulcers.

Navigate to figure in ReviewOpen in new tab

Analysis 1.4

Comparison 1 Electrical stimulation (plus standard care) versus sham/no electrical stimulation (plus standard care), Outcome 4 Time to complete healing.

Navigate to figure in ReviewOpen in new tab

Analysis 1.5

Comparison 1 Electrical stimulation (plus standard care) versus sham/no electrical stimulation (plus standard care), Outcome 5 Rate of pressure ulcer healing.

Navigate to figure in ReviewOpen in new tab

Summary of findings for the main comparison. Electrical stimulation (plus standard care) versus sham/no ES (plus standard care) for treating pressure ulcers

Electrical stimulation (plus standard care) versus sham/no ES (plus standard care) for treating pressure ulcers
Patient or population: people with pressure ulcers Setting: inpatients and outpatients Intervention: electrical stimulation (plus standard care) Comparison: sham/no ES (plus standard care)
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	Number of ulcers (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with sham/no ES (plus standard care)	Risk with Electrical stimulation (plus standard care)
Proportion of pressure ulcers healed (3 to 12 weeks)	Study population		RR 1.99 (1.39 to 2.85)	512 (11 RCTs)	⊕⊕⊕⊝ Moderate^a	ES may increase the proportion of pressure ulcers healed when compared with no ES. Absolute effect: 297 out of 1000 (from 207 more to 425 more).
	149 per 1,000	297 per 1,000 (207 to 425)
Time to complete healing (3 and 8 weeks)	Study population		HR 1.06 (0.47 to 2.41)	55 (2 RCTs)	⊕⊝⊝⊝ Very low^b	It is uncertain if ES decreases time to complete healing when compared with no ES.
	18 per 100	19 per 100 (9 to 38)
Complications/ adverse events related to pressure ulcers (3 to 12 weeks)	Adverse events included redness of the skin, itchy skin, dizziness and delusions, deterioration of the pressure ulcer, limb amputation and occasionally death.			602 (13 RCTs)	⊕⊕⊝⊝ Low^c	The data were not sufficiently detailed or comparable to analyse quantitatively.
Quality of life (QoL)	No studies measured quality of life
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; RR: Risk ratio; OR: Odds ratio;
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: we have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect
^aDowngraded one level: once for serious risk of bias because a lot of the studies had either high or unclear risk of bias for performance bias and selective reporting. ^bDowngraded four levels: once for serious risk of bias because both studies had high risk of bias for two domains and one study had unclear risk of bias for another three domains; once for indirectness because the two studies were not reflective of all who are vulnerable to pressure ulcers; twice for imprecision. ^cDowngraded two levels: once for serious risk of bias because a lot of the studies had either high or unclear risk of bias for selection and attrition bias; once for imprecision.

Summary of findings for the main comparison. Electrical stimulation (plus standard care) versus sham/no ES (plus standard care) for treating pressure ulcers

Navigate to table in Review

Table 1. Sensitivity analyses‐ bias

Outcomes

Pooled results

Randomisation (studies with
adequate sequence generation)

Allocation (studies with
concealed allocation)

Assessors (studies with
blinded assessors)

Dropout rate (studies
with < 15% dropouts)

Proportion of

pressure ulcers healed

1.99 (1.39 to 2.85)

(n = 12)

2.12 (1.36 to 3.30)

(n = 6)

1.98 (1.35 to 2.90)

(n = 7)

1.93 (1.26 to 2.95)

(n = 9)

2.34 (1.47 to 3.71)

(n = 6)

Results are presented as RR (95% CI)
n = number of studies included in analysis

Table 1. Sensitivity analyses‐ bias

Navigate to table in Review

Comparison 1. Electrical stimulation (plus standard care) versus sham/no electrical stimulation (plus standard care)

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Proportion of pressure ulcers healed Show forest plot	11	512	Risk Ratio (IV, Random, 95% CI)	1.99 [1.39, 2.85]

2 Composite measure of pressure ulcer severity Show forest plot	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

3 Surface area of pressure ulcers Show forest plot	12		Mean Difference (IV, Random, 95% CI)	Totals not selected

4 Time to complete healing Show forest plot	2	55	Hazard Ratio (Fixed, 95% CI)	1.06 [0.47, 2.41]

5 Rate of pressure ulcer healing Show forest plot	12	613	Mean Difference (IV, Fixed, 95% CI)	4.59 [3.49, 5.69]

Comparison 1. Electrical stimulation (plus standard care) versus sham/no electrical stimulation (plus standard care)

Navigate to table in Review

Cochrane Review language

Website language

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

PICOs

PICOs

Population

Intervention

Comparison

Outcome

Plain language summary

Is electrical stimulation effective for treating pressure ulcers?

Visual summary

Authors' conclusions

Implications for practice

Implications for research

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Searching reference lists of included trials and relevant reviews

Searching by contacting individuals or organisations

Adverse effects

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Cross‐over studies

Studies with more than one ES group

Studies in which multiple observations were taken on the same individual

Studies in which more than one pressure ulcer was treated per participant

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

'Summary of findings' tables

Ethics and inequalities

Context

Results

Description of studies

Results of the search

Included studies

Study design and setting

Participants

Interventions

Outcomes

Registry and funding source

Excluded studies

Studies awaiting classification

Ongoing studies

Risk of bias in included studies

Allocation

Random sequence generation

Concealed allocation

Blinding

Blinding of participants