Article Text

PDF

Pilot testing of methods for evaluation of acupuncture for emesis during radiotherapy: a randomised single subject experimental design
  1. Anna Enblom1,2,
  2. Annica Tomasson3,
  3. Mats Hammar4,
  4. Gunnar Steineck5,6,
  5. Sussanne Börjeson1,7
  1. 1Department of Medical and Health Sciences, Division of Nursing Science, Linköping University, Linköping, Sweden
  2. 2The Vårdal Institute, Sweden
  3. 3Rörelse och Hälsa, Primary Health Care, County Council of Östergötland, Linköping, Sweden
  4. 4Department of Clinical and Experimental Medicine, Division of Obstetrics and Gynecology, Linköping University, Linköping, Sweden
  5. 5Department of Oncology, Division of Clinical Cancer Epidemiology, Karoliniska Institute, Stockholm, Sweden
  6. 6Department of Oncology, Division of Clinical Cancer Epidemiology, Sahlgrenska Academy, Gothenburg
  7. 7Centre of Surgery and Oncology, Department of Oncology, Linköping University Hospital, Linköping, Sweden
  1. Correspondence to Dr Anna Enblom, IMH omvårdnad, c/o S. Börjeson BSP plan 12, Hälsouniversitetet, SE-581 85 Linköping, Sweden; anna.enblom{at}ki.se

Abstract

Background Many acupuncture studies are of weak methodological quality, possibly due to lack of pilot testing. This pilot study tested design features, including test of feasibility, compliance to treatment and data collection, level of blinding success and the patients' potential perceived effects of the treatment, in preparation for an efficacy study.

Method A modified single subject experimental design was conducted. 10 cancer patients were randomised to verum penetrating acupuncture or non-penetrating sham needles for 30 min 2–3 times/week during radiotherapy over abdomen/pelvis. They answered test-retested emesis questions (r=0.527–1.0) covering nausea, vomiting, use of antiemetics, wellbeing and activities of daily living.

Results Overall, the patients completed 98% of the 345 emesis-questionnaire days and 101 of the 115 offered treatments. All patients believed they received verum acupuncture. 10 patients experienced antiemetic effects, seven relaxation, five pain-reduction and five experienced sleep improvement. Two types of nausea questions showed absolute concordance (r=1.0) (n of observations=456). Nausea was experienced by one of five verum acupuncture treated patients (duration median 0% of the radiotherapy-days) and four of five sham acupuncture treated patients (duration median 24% of the radiotherapy-days). Patients experiencing nausea rated decreased wellbeing and performance of daily activities compared to patients free from nausea.

Conclusions All patients were blinded, complied with verum/sham treatments and data-collection, and believed they had effects of the received treatment. The methods for verum/sham treatment and data collection may thus be used in an adequately powered randomised controlled study of the effect of acupuncture for radiotherapy-induced emesis.

Statistics from Altmetric.com

Introduction

It is important to create a relevant study design for testing the effect of acupuncture on emesis (nausea and vomiting) in cancer patients undergoing radiotherapy. There are no previous studies in this area, though several on chemotherapy-induced emesis. They show few side effects but report conflicting results regarding effects on emesis.1 The studies all have some methodological problems—for example, lack of control groups,2 3 lack of sham treated controls,4,,6 controls treated with penetrating acupuncture,7 8 blinding problems9 or lack of statistical power.10 In light of the large number of acupuncture studies suffering from methodological problems, it seems important to describe an example of a thorough pilot testing of acupuncture design features. It seems obvious that all acupuncture studies should have a pilot phase, but many antiemetic acupuncture studies had a weak design,1 indicating that pilot studies had not identified eventual weaknesses.

Summary points

  • The effect of acupuncture on radiotherapy-induced nausea is not known.

  • We conducted a pilot study using the novel single subject experimental design.

  • Patients were successfully blinded, complied with treatments and data collection, and believed they had antiemetic effects of treatment.

Both the feasibility and the potential effects of acupuncture on radiotherapy-induced emesis, which continues for weeks,11 may differ from those on chemotherapy-induced emesis, treated most often 1–5 days after the chemotherapy.1 In all, 60–70% of patients irradiated over abdomen or pelvis experienced nausea.12,,14 Emesis starts, often delayed after several daily fractions, then continues for several weeks during and after the end of radiotherapy,11 which may reduce wellbeing12 15 16 and performance of daily activities.12 17 Antiemetics, especially serotonin-receptor antagonists are effective.14 However, our previous study including 368 patients irradiated to a variety of fields,11 indicated that radiotherapy-induced emesis is often not treated adequately. One third of the nauseous patients considered their antiemetic treatment insufficient and only a quarter thought that antiemetics had helped them at least moderately. Three-quarters of all patients were interested in receiving acupuncture in the future for nausea or in receiving more information about acupuncture. In oncology care in general, many cancer patients18 and oncology healthcare professionals19 are interested in non-pharmacological complements to medical treatment but we do not know if acupuncture is a feasible treatment during radiotherapy.

Aim and specific objectives

The aim of this pilot study was to test specific design features in preparation for a large randomised controlled trial evaluating the effect of acupuncture on radiotherapy-induced nausea. The objectives were to investigate the feasibility of performing verum or sham acupuncture treatment two to three times per week during the radiotherapy period, compliance to treatment and data collection, the level of blinding success, and the patients' potential perceived effects of the treatment. Our study highlighted the individual patient situation during verum or sham acupuncture treatment during radiotherapy; regarding emesis, use of antiemetics and self-care treatment, quality of life and performance of daily activities.

Method

The study used a case-control single subject experimental design (SSED),20 following criteria for a SSED21: frequent and repeated measures of the dependent variable throughout each phase using precisely defined methods (replicable and reliable), careful control of the manipulation of the treatment variable (one treatment variable manipulated at a time) and analysis of its effect on the outcome variables, clearly specified and operationally defined target behaviours and control of extraneous variables. Additional recommendations were followed: independent (blinded) evaluator, and visual and statistical analyses of data.21 In a classically conducted SSED (eg, using an A1BA2 design),20 21 patients with stable symptoms (eg, chronic pain) are randomly allocated to alternating receive treatment (intervention=B-phase) or no treatment (baseline=A-phase). In our particular setting, the SSED was modified by using an A1BCA2 design to meet the demands that the expected natural emesis variation require. We followed each of the ten participating patients separately during the study phases: baseline within 4 weeks before intervention start (A1-phase), intervention during the whole radiotherapy period (B-phase), follow-up during 2 weeks after radiotherapy (C-phase), and a second baseline 4 weeks after the end of radiotherapy (A2-phase). Another modification of a classically conducted SSED20 21 was made: a nurse randomised the patients to intervention (verum acupuncture: five cases) or control (sham acupuncture: five controls) during the B-phase, by drawing lots from a closed envelope. The study was approved by the regional ethics committee and followed the guidelines for ethical complementary research22 and for acupuncture treatment in cancer patients.23 We used the Standards for Reporting Controlled Trials of Acupuncture (STRICTA) checklist 24 in the planning and reporting of the pilot study.

Participants

During a 6 month period we included all patients at least 18 years old, admitted to the oncology department of a Swedish University Hospital to receive radiotherapy over abdominal or pelvic fields of at least 25 Gray dose for gynaecological, testicular, anal or rectal cancer tumours, who were able to take part in the whole study procedure and gave informed consent. Exclusion criteria were antiemetic treatment or frequent (almost daily) persistent nausea during the past month, present during the 24 h prior to the start of radiotherapy; previous acupuncture treatment for nausea; or acupuncture treatment within the past year regardless of indication.

Patients were informed orally and in writing that they would receive—without being told which—either the usual type of acupuncture with needles penetrating the skin or treatment with needles placed just against the skin and that the effect of either treatment type was unknown.

Radiotherapy and antiemetic treatment

The patients were irradiated once per day, Mondays through Fridays, during the B phase. The oncologists, blinded to acupuncture treatment allocation, prescribed rescue antiemetics according to the standardised normal clinical routines at doses based on The Swedish Medicine Information Engine (http://www.fass.se). If nausea occurred, the oncologists prescribed an antihistamine with antiemetic effect (meclozine 25 mg 1–2 times/day). If that did not give satisfactory effect, they added a serotonin receptor-antagonist (ondansetron 8 mg, twice/day, Granisetron 2 mg, once/day, or tropisetron 5 mg, once/day). If that did not give satisfactory effect, they also added a corticosteroid (betamethasone 4 mg, twice/day). The antiemetic consumption is described in figure 1.

Figure 1

Nausea and vomiting during and after the radiotherapy period in patients treated with verum acupuncture (upper graph) and sham acupuncture (lower graph). Phase B: intervention phase during radiotherapy. Phase C: follow-up during 2 weeks after the end of radiotherapy. Phase A2: second baseline at 4 weeks after the end of radiotherapy. Numbers of days with V (vomiting), D (dopamine receptor antagonists), S (serotonin receptor antagonists) and C (corticosteroids) are presented beside the patients' lines. *Patient 1 and 6 received antiemetics because of chemotherapy.

Three unblinded physiotherapists started verum and sham treatments in association with the first radiotherapy occasion, before the possible onset of emesis, and repeated treatment for 30 min three times per week for the first 2 weeks, then twice weekly during the remaining radiotherapy period. Both verum and sham acupuncture needles were steel needles packed sterile, manufactured by Dong Bang Acupuncture(EU)LTDv, Korea. The therapists had education in acupuncture (comparable with 10 full-time weeks of university studies). The numbers of years with experience as physiotherapists/acupuncturists were: 4/3, 7/3 and 26/11, respectively. The physiotherapists followed a standardised treatment protocol covering routines for placing and stimulation of the needles and conversation with the patients (allowing everyday conversations, avoiding the subject emesis).

Verum acupuncture

Verum acupuncture was administered bilaterally to the standard antiemetic point PC61,,3 using needles, diameter 0.30 × length 40 mm, inserted to a depth of a half body-inch. The physiotherapists manually induced ‘de qi’ three times per session by twirling and lifting. ‘De qi’ was reached when the patient reported a sense of numbness and soreness and the physiotherapist noted a minimal muscular contraction around the needle.25

Sham acupuncture

The therapists used the telescopic, non-penetrating Park's sham needles26 for the controls. The credible27,,29 sham needle, 0.30 × 40 mm (fully extended length) was placed bilaterally at a non-acupuncture point twice the distance of PC6 from the wrist. The sham needle looks identical to a verum acupuncture needle but when the blunt sham needle touches the skin and gives a sensation of penetration, it glides upwards in its handle without being inserted into the tissue at all. The therapists simulated needle stimulation by twirling the blunt needles against the skin for a couple of seconds three times per session, not producing de qi.25 Therefore the sham treatment was different from both acupuncture and acupressure therapy. Park's guide tube, the bottom of which was covered with double sided tape, fixed the sham needles to the skin. The tubes marked the needling points in both cases and controls so that the both procedures looked identical.

Data collection

A nurse, not involved in the acupuncture treatment, distributed coded questionnaires to the patients who answered them alone and sent them to the study evaluator (one of the authors). The evaluator was blinded to treatment allocation until after analysis of the emesis results. The study-specific parts of the data collection were tested for face validity in nine radiotherapy patients and for test-retest reliability in 36 irradiation/chemotherapy patients, resulting in the Spearman's correlations shown in table 1.

Table 1

Examples of questions included in the emesis questionnaire

The phases A1BCA2

In phase A1 (baseline) the patients detailed socio-demographic data, experiences of nausea in previous situations, and medical record data were collected (table 2). The patients also answered two study-specific single-item questions, covering wellbeing and performance in daily activities (table 1), and the Swedish version30 of the widely used, valid and reliable Functional Assessment of Cancer Therapy-General (FACT-G).31 32 The instrument consists of 27 statements covering the preceding week, answered from 0=‘do not agree at all’ to 5=‘totally agree’. Higher scores represent higher health-related quality of life.

Table 2

Characteristics of the patients

During the B-phase (intervention), each morning the patients answered a study specific emesis-questionnaire33 34 covering nausea, vomiting and antiemetic consumption within the preceding 24 h period (table 1). Every seventh morning, the patients answered questions regarding wellbeing and performance of daily activities (table 1) and stated if they had experienced positive effects on relaxing, mood, sleep and pain (‘yes’ or ‘no’) during or close to the treatment sessions.

The treating physiotherapist inspected and asked the patients about possible side effects immediately after every acupuncture/sham session. At the first, the sixth and the last session the patients answered the question: ‘Do you think that the treatment that you just received is effective to prevent and reduce nausea?’ (‘No, I do not think the treatment is effective’ or ‘Yes, I believe a little/moderately/much that the treatment is effective’). After the last treatment the patients also answered the question: ‘Do you think you have been treated with needles that have penetrated the skin, or do you think the needles have been placed just against the surface of your skin?’ and answered the FACT-G again.

In phase C (the follow-up), the patients during 2 weeks after the end of radiotherapy continued to complete the daily emesis-questionnaire and the weekly questions regarding wellbeing and daily activities (table 1).

In phase A2 (second baseline), the patients answered the emesis-questionnaire, the questions regarding wellbeing and daily activities (table 1), FACT-G, and the question: ‘If you in the future would need a similar radiotherapy, would you then be interested in receiving acupuncture for nausea?’ (‘No I am not’, ‘Yes, a little’, ‘Moderately’, ‘Much’ interested).

Analyses

The expected natural emesis variations during radiotherapy,11 without acupuncture would be: no emesis at baseline (phase A1), nausea in approximately 60% of patients12 13 after several fractions of radiotherapy (phase B), end of nausea within the first 2 weeks after radiotherapy (phase C) and no nausea at 4 weeks after radiotherapy (phase A1). If acupuncture treatment was completely successful, all phases would be free from emesis. The data were, according to methods often used in SSED-studies,20 21 as a first step shown separately for each patient, allowing visual assessment of changes within and across the patients (figure 1, table 3 and table 4). The repetitive measurements used in a SSED mean that the data points are not independent.35 Therefore, as a second step, the autocorrelation test35 was used to evaluate if series dependence existed between the measurements point regarding nausea. There was no statistically significant autocorrelation for any nauseous patient in the B-phase. Thus, as a third step, a visually assessed change of one step or more in nausea ratings from phase A1 (zero ratings), to phase B or A2, and of one step or more from phase B to phase C was, according to previous findings,15 considered to be a clinical meaningful change. Proportion of radiotherapy days with nausea and vomiting and completion of treatments and questionnaire answers (performed number divided by planned number) was calculated for each patient. The median (Interquartile range; IQR) proportion of nausea and vomiting days for the patients was presented by group.

Table 3

Compliance to treatment and to questionnaires

Table 4

Emesis and experienced positive effects of the treatment

The content validity of the emesis-questionnaire was calculated: concordance between answers regarding nausea occurrence measured with the two-category question (yes/no) and with respect to intensity (dichotomised into ‘not relevant’=no, and ‘little’ to ‘very severe’=yes) was calculated by Spearman's correlation coefficient (r) (table 1).

Results

Characteristics of the patients

Nineteen patients were screened for eligibility. Nine could not be included (eight did not want to participate and one consumed antiemetics before the start of radiotherapy). Eight women and two men participated in the study. Both the men were randomised to sham and the median age was lower (60 years) in the patients treated by sham than in the patients treated by verum acupuncture (75 years). All patients treated with sham had seldom previously experienced nausea, while four patients treated with verum acupuncture had experienced nausea every 6-months or more often (table 2).

Blinding, compliance and quality of the study design

All patients continued verum or sham acupuncture during their whole radiotherapy period and believed they received penetrating acupuncture. The patients missed a median of one single treatment (range 0–4). They answered a total of 474 questionnaires out of the planned 505 (94%) (table 3). There was an absolute concordance (r=1.0) between answers to the two types of nausea questions (number of ratings with each type of question was 456 during B to C phase). The sham needle scratched the patients' frail skin once (patient 6 and 2), patients 6 and 9 experienced minor dizziness and patient 2 felt tired after one session.

Emesis, consumption of antiemetics and self-care for emesis

Figure 1 shows the changes in emesis and antiemetic consumption within and across the patients. Nausea and vomiting within the radiotherapy period was experienced by one (patient 9) of the five cases receiving verum acupuncture and nausea by four (patient 2, 3, 6 and 8) and vomiting by two (patient 3 and 6) of the five controls receiving sham acupuncture. As shown in table 4, the median proportion of radiotherapy days with nausea was 0% (IQR 0–0%) in the cases receiving verum acupuncture and 24% (IQR 10–30%) in the controls. Two patients used self-care to reduce emesis: modified eating habits (patient 6 for 6 days, patient 9 for 5 days), increased intake of salt (patient 6 for 1 day), drank healthy drinks (patient 6 for 1 day, patient 9 for 7 days), ate products with healthy abdominal bacteria culture (patient 9 for 4 days) and ate dry biscuits (patient 9 for 3 days).

Subjectively experienced effects

All patients, except patient 3, believed at least moderately that their treatment was effective for nausea. Eight patients stated they also experienced other positive effects and were interested in receiving the treatment again for nausea (table 4).

Quality of life, wellbeing and activities in daily living

The patients rated FACT-G stable at the end of the radiotherapy period (phase B; median 47) compared to baseline (phase A1; median 49). At the 4-week follow-up, all patients rated an increase (phase A2; median 73). The patients who experienced nausea rated lower wellbeing and proportion of performed daily activities (median ‘little’ and ‘fewer than half of daily activities’) compared to the patients who were free from nausea (median ‘moderate’ and ‘more than half of daily activities’).

Discussion

In this pilot study, patients were successfully blinded, complied with treatments and data collection and believed they had antiemetic and other treatment effects.

A classical SSED study does not include randomised treatment allocation,20 but it seemed important to use randomisation to test the blinding of patients. Similarly, Park and co-workers,28 found that none of 45 patients treated by verum acupuncture patients and only four of 49 patients treated with the telescopic sham needle thought they had received sham. The telescopic sham needles,26 29 have been considered a major advance in acupuncture research24 but were used in only four out of 47 acupuncture studies, according to one review.36 In our study, the therapists were not blinded, though might have been blinded by using the sham needle developed37 and validated38 by Takakura and co-workers. However, in our study, the unblinded therapists probably did not influence emesis data differently between the groups since the therapists were trained in performing credible sham acupuncture18 and used a strictly standardised treatment protocol during the treatments. The evaluator and all other healthcare professionals were blinded.

Needles inserted superficially,1 9 or deeply at non-acupuncture points,7 8 frequently used in previous antiemetic studies cause greater sensory stimulation than a blunt needle.25 39 However, the telescopic sham needles are not totally physiologically inactive and may activate the limbic system.40 Pariente and co-workers39 found that the blunt needle induced activity in the mid-brain, which was related to expectations rather than the blunt needle itself. In our study, we used a non-acupuncture point to avoid a hypothetical acupressure effect on PC6. Further, the duration of needle pressure on the skin was minimal (total of approximately 100 s during the median of 10 treatments per patient).

Treating the PC6 point bilaterally for approximately 30 min is a treatment often used,6 9 10 though some only used 5 min7 and sometimes only one single treatment.4 8 10 In the present pilot study, the patients complied with treatment during the long period of radiotherapy, in that no patient dropped out and few treatments were missed. The patients completed 98% of the planned questionnaires during the radiotherapy treatment period. The high compliance and lack of major side effects indicate that the study procedure was acceptable for patients and could be used in a larger study of acupuncture. However, the response rate did decrease after the end of radiotherapy, possibly due to less contact with the healthcare professionals, indicating the importance of sending reminders in future studies. The high compliance with data collection may be due to the use of face-validated study-specific questionnaires; similar figures have been shown by others regarding cancer related issues using study-specific questionnaires in Scandinavia.12 41

Test-retest validity of the emesis questionnaire after a 1 h interval, in 36 irradiation/chemotherapy patients was very high. We chose the short interval of 1 h to ensure the actual perceived severity of nausea was constant. The study-specific emesis-questionnaire includes methods for measuring nausea (two-category and five-category scales) separately from vomiting, as recommended,42 used12 15 and validated.33 34 There are alternative emesis scales with more items,43 probably too bothersome for daily use in these patients. Single-item questions about wellbeing and activities of daily living reduced patient burden and provided satisfactory predictive validity, discriminating nauseous patients from those free from nausea.

Visual inspection of emesis results indicated a reduction of the proportion of emesis days in patients receiving verum acupuncture compared to patients receiving sham acupuncture, though no between-group statistical analysis is presented for lack of power. The suggested reduction of emesis in the verum acupuncture group may be a result of acupuncture or of confounding factors differently distributed between patients. Despite randomisation, both young men received sham while patients with larger experience of nausea received verum acupuncture. Male gender decreases the emesis risk, while an age below 40 years, previous experience of nausea, and a testicular tumour increases the emesis risk.11,,13 We excluded patients with significant nausea before radiotherapy to minimise the threats against the internal validity by avoiding changes in other sources of nausea, for example, vestibular neuritis. In a larger randomised study, confounding factors for nausea would be equally distributed between groups.

Though studies of acupuncture for radiotherapy-induced emesis are rare, there are several in chemotherapy-induced emesis.1 Patients receiving manual or electroacupuncture, or acupressure experienced less acute emesis than patients receiving only antiemetics.1 Daily electroacupuncture reduced delayed emesis more than superficial needles with sham electrostimulation9 but one single manual verum acupuncture treatment did not reduce acute nausea more than telescopic sham needles.10 Our study is compatible and patients were interested in receiving future treatment. Satisfaction with acupuncture was in concordance with other studies in which 89% thought it was very important that the clinic continued to offer acupuncture for various cancer-related symptoms2 and 93% believed that acupuncture was helpful in reducing chemotherapy-induced emesis.3

A SSED has limited generalisability,20 21 44 especially when patients are consecutively rather than randomly selected and in a single setting. This SSED will be followed by a standard randomised controlled multi-centre trial. An A1BCA2 design was considered optimal since the repeated A-phase controls for the natural history. Thus, the C and A2-phases are strengths in the study; using an AB design only is the weakest of the SSE-designs because of its inability to control for potential threats of the internal validity. A limitation of our study was the need to modify SSE-design. It is common to use multiple baseline phases during the time when the patient is experiencing the symptom under study,20 21 to enhance control of the effect of confounding events. A multiple baseline design model classically consists of an A1BA2 design with different duration of baseline A1.44 This would mean all phases should have been performed during radiotherapy. That classical A1BA2 design requires stability of the dependent variable within each phase.45 SSED studies have mostly been performed in stable conditions—for example, during neurological46 or chronic pain47 rehabilitation. Our patients receiving radiotherapy had fluctuating or increasing emesis.11 Our modifications, with A1 before radiotherapy and A2 afterwards, ensured stability during these phases. However, the dependent variable was not stable within phase B, leading to the second modification to classical SSE design: randomisation to one treatment during the whole radiotherapy period. It was not appropriate to use alternative treatments or withdrawal design (treatment followed by no treatment)20; it seemed unethical and would have affected the next B-phase, due to the risk for anticipatory nausea.48 The fact that we used a randomised controlled design during the B-phase allowed pilot-testing of blinding and allowed control for caring or expectation-effects. The continuous delivery of radiotherapy also hindered any manipulation of the timing of phases change (eg, acupuncture applied to one patient while the other patients continue baseline); acupuncture is supposed to be provided before start of emesis.1 Except from these modifications, our study followed all regular and two additional (independent, blinded evaluator and visual and statistical analyses of data) criteria for SSED.21 The C-phase allowed data to be stable and controlled for the natural history. In SSED studies, the visually inspected changes from A to B-phases are often analysed statistically, by using, for example, the split middle procedure, the Rn statistic or the two SD band method.20 21 35 These methods were not applicable for our data (ordinal), so we used a change of one step or more in the rating.11

We followed the STRICTA checklist in our report.24 though most acupuncture studies with cancer patients did not.45 We believe a pilot study using SSED,20 21 reported according to the STRICTA checklist,24 may provide valuable information regarding the study design and feasibility, especially in patients with chronic or stable symptoms. In this pilot study, all patients were blinded, complied with treatments and data collection, and believed they had antiemetic effect of the received treatment. Thus, we consider the present methods for verum/sham treatments and data collection to be relevant for a randomised controlled study of the effect of acupuncture for radiotherapy-induced emesis, as a complement to potent antiemetics.

Acknowledgments

The authors thank the participating patients. They are also thankful to Boel Lindberg and Ingrid Tillgren for coordinating, Eva Ahlner for treatments, Beata Bergius-Axelsson for pretesting acupuncture therapy and Anna Sjöberg and Amelie Svejdur for pretesting emesis questionnaires.

References

View Abstract

Footnotes

  • Funding This study was supported by The Vårdal Foundation—for Health Care Sciences and Allergy Research, The Vårdal Institute, The County Council of Östergötland and The University of Linköping.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the Regional Ethics Committee Linköping University, Dnr 02-420.

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

Request permissions

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