Puin, 44 (1991) 221-229 8~ 1991 Elsevier Science Publishers ADONIS 0304395991~~1 L

221 B.V. 03~-3959/9~/~03.50

PAIN 01734

An in-depth study of long-term users of transcutaneous electrical nerve stimulation (TENS). Implications for clinical use of TENS MI. Clinical P~vchopharmacoiogV

Johnson,

Unit, Department

(Received

C.H.

Ashton

and J.W. Thompson

of Pharmacological Scxnces, The Medrcal School, Unicersit,v Newcastle upon Tyne NE2 4HH (U.K.)

5 March

1990, revision

received and accepted

17 August

of .~rewctisife upon Qne.

1990)

This in-depth study examines the relationships between patient, stimulator and outcome variables in Summary a large number of chronic pain patients utilising TENS on a long-term basis. 179 patients completed a TENS questionnaire designed to record age, sex, cause and site of pain and TENS treatment regime. Of these 179 patients, 107 attended our research unit for assessment of the electrical characteristics of TENS during self-administered treatment. Although a remarkable lack of correlation between patient, stimulator and outcome variables was found to exist, the analysis revealed much information of importance: 47% of patients found TENS reduced their pain by more than half; TENS analgesia was rapid both in onset (less than 0.5 h in 75% patients) and in offset (less than 0.5 h in 51% patients); one-third of patients utilised TENS for over 61 h/week; pulse frequencies between 1 and 70 Hz were utilised by 75% of patients; 44% of patients benefitted from burst mode stimulation. The clinical implications of these findings are discussed. Key words: Transcutaneous efficacy

electrical

nerve stimulation

Introduction When TENS is used to treat a patient’s pain, a number of important variables related to the patient and to the stimulator may influence treatment outcome. In this study, for the first time, the relationship between patient variables (e.g., age, sex, cause and site of pain, personality and use of drugs), stimulator variables (e.g., model of stimulator, site of electrodes, pulse waveform, pulse frequency (Hz), pulse pattern, pulse intensity (mA current) and pulse width (uses)) and outcome vuriab~e~ {e.g., TENS analgesic efficacy, onset of analgesia, postTENS analgesia and adverse effects) have been examined in depth in a large number of patients. Linzer and Long [12] were able to monitor the electrical characteristics and electrode positions in only 23

Correspondence fo: Prof. J.W. Thompson, Clinical Psychopharmacology Unit, Department of Pharmacological Sciences, The Medical School, University of Newcastie upon Tyne, Framlington Place, Newcastle upon Tyne NE2 4HH, U.K.

(TENS);

Analgesia;

Chronic

pain

patients;

TENS

clinical

patients suffering from chronic pain who obtained satisfactory TENS analgesia. They concluded that the success rate with TENS could be improved by careful attention to electrode placement, stimulating variables and patient education. Many studies followed in which stimulating characteristics were examined. Thus, Mannheimer and Carlsson [ 131 found that conventional TENS delivered at a frequency of 3 Hz was less effective in suppressing pain than conventional TENS at 70 Hz for patients with rheumatoid arthritis. However ‘bursting’ trains of pulses at 3 Hz (acupuncture-like TENS) as described by Eriksson et al. [4] also proved beneficial for these patients. Wolf et al. [23] attempted to determine optimal electrode placements and stimulating parameters while treating chronic pain patients with conventional TENS, administered at the clinic in 30-45 min sessions. No clear correlations between electrode placements, stimulating parameters and pain relief were found, although TENS did reduce pain intensity scores. The group concluded that a patient’s psychological profile may help to determine who will benefit from TENS. Bates and Nathan [2] summarised the situation ‘after 7

222

years familiarity with transcutaneous stimulation, the present authors are still unable to predict whether or not a particular patient will obtain long lasting pain relief.’ In practice. conditions for stimulation of individual patients are usually found by trial and error. Despite its obvious clinical and therapeutic value, information on the treatment regimes achieved by long-term users of TENS is limited. The aim of the present study was to examine the relationships between patient, stimulator and outcome variables. A total of 179 randomly selected patients with chronic pain who were currently using TENS to control their condition completed a specially devised TENS multiple choice questionnaire (TENS MCQ). Of these patients, 107 also attended our research unit where the electrical characteristics of stimulation used by the patient for therapy were recorded. The results of this survey are presented as histograms and the implications for future treatment regimes discussed.

Method

Stud,, populution One hundred and seventy-nine patients (female n = 82, male n = 97; age range 24-85, mean + S.D. = 55.2 k 12.9) with a variety of chronic pain conditions drawn randomly from our files participated in the study. All had been in possession of a TENS stimulator for at least 3 months (loaned to them by Newcastle Pain Relief Clinic after a successful trial session). All patients completed a TENS multiple choice questionnaire (TENS MCQ) designed by the authors. The questionnaire collected data regarding site and cause of pain, the model of stimulator. frequency of stimulator use, degree of pain relief and current treatment. Personality variables were also measured using the Eysenck Personality Questionnaire (EPQ) [6]. All patients were classified in 3 ways: (i) according to the main anatomical region of pain and (ii) on the basis of aetiology of pain, both conforming to the Classification of Chronic Pain prepared by the International Association for the Study of Pain (IASP [8], this reference gives a full description of the classification); (iii) a third group termed ‘diagnostic clusters’ was created by the authors and included patients clustering into miscellaneous diagnostic categories. Determinution UlXl~).slS

of electrical charucteristics

of TENS;

Frye

Of the 179 patients, 107 (female n = 58, male n = 49; age range 24-85) attended the research unit so that the electrical characteristics of TENS could be assessed (termed ’ Frye analysis’). The electrical characteristics of

TENS were assessed by recording peak voltage (V), pulse width (psec). pulse frequency (Hz), pulse intensity (mA current) and electrode impedance (kfJ) on a Type 4000 Frye analyser (RDG Electra-Medical). Patients applied TENS to the ‘usual’ anatomical site to achieve analgesia for their particular chronic pain condition. They were asked to adjust both pulse intensity and pulse frequency (via the appropriate control dials) to the settings they normally used to treat their pain. Patients free of pain during the visit were asked to adjust the stimulator to its ‘normal therapy setting.’ Electrical characteristics were recorded at sensory threshold, therapy level and pain threshold to TENS electrical pulses at the site of electrode application. Patients found respective thresholds after a small amount of ‘hunting’ and the mean of 3 repetitions was calculated. Recordings were taken at the patient’s preferred frequency of pulse delivery (Hz) and repeated at high frequency (100 Hz) and low frequency (20 Hz) stimulation. If stimulators possessed a burst facility. the entire procedure was repeated in the burst mode. Patients were in possession of one of the following types of stimulator: (1) Tiger Pulse (polarity facility, RDG Medical). (2) Tiger Burst (burst facility, RDG Medical). (3) Spembly 9000 (burst facility, Spembly). (4) Microtens 7757 (burst facility, Neen Pain Management Systems).

Twenty-four patients (female n = 12. male n = 12; age range = 39-79. mean + S.D. = 54.2 f 9.8) returned for a follow-up assessment. A 4 month period elapsed between the two visits.

Results

A wide range of chronic pain conditions are represented in the patient population. One hundred and seventy-seven patients were classified and distributed into the following categories according to: (i) Anutomicul region (IASP axis i); head 4.5% (8) cervical 3.9% (7) upper limbs 10.2% (18). thoracic 12.4% (22). abdominal 9.6% (17). lower back 37.8% (67) lower limbs 14.1% (25). pelvic 0.0% (0). anal 0.6% (1). 2 or more sites of pain 6.8% (12). (ii) Aetiology (IASP axis v); congenital 1.1% (2) trauma 23.7% (42) infective 9.0% (16). inflammatory 2.8% (5), neoplasm 0.0% (0) toxic 3.9% (7). degenerative/ mechanical 44.1% (78) dysfunctional 7.3% (13). other 3.4% (6) psychological 4.5% (8). (iii) Diqnostic clusters (authors classification); deafferentation 15.7% (27) entrapment 31.4% (54), neuralgia 13.9% (24). spondylosis 13.9% (24) myofascial

223

9.3% (16) sympathetic 6.4% (11). Raynaud’s other 6.4% (11). psychological 4.6% (8).

1.2% (2).

Questionnaires

15 (0

The results of the 179 questionnaires are summarised in Table I. As patients were instructed to omit any questions that they did not understand, the n values varied for different results. Thirty-seven per cent of patients (67) were using a Microtens 7757, 28% (49) a Tiger Burst, 25% (44) a Tiger Pulse and 10% (17) a Spembly 9000. The analgesic efficacy of TENS assessed by a visual analogue scale (VAS; where 0 equates to zero relief of pain and 10 equates to total relief of pain) is shown in Fig. 1. Forty-seven per cent of patients (79) found TENS reduced their pain by half or more; 13.7% (23) of patients reported that TENS did not produce any relief (between 0 and 1 VAS) and 15.5% (26) achieved total relief of pain (between 9.1 and 10 VAS) using TENS. A clustering effect was observed around the columns representing total relief of pain (VAS = 10). no relief of pain (VAS = 0) and relief of pain by half (VAS = 5).

TABLE

I

SUMMARY OF ANSWERS QUESTIONNAIRE

TO THE TENS

MULTIPLE

CHOICE

Question

”

Mean (range)

Age of study population

(a) 82 (b) 97 (c) 179 179 168

(a) 55.7 (24-82) (b)54.8 (30-85) (c) 55.2 (24-85) 4.0 (0.25-9) 5.1 (O-10)

(a) Female (b) Male (c) All (years) Length of use (years) Degree of pain rehef (0 = none H 10 = total) Duration of treatment (h/week) Time to onset of analgesia (min) Duration of post-TENS analgesia (min)

152 150

39.7 30

(0.75-63) (O- >2 h)

150

51

(0- z 2 h)

” Pulse pattern preference

(a) continuous (b) burst (c) both Pulse frequency (a) fast preference (b) slow (c) none Regularly reset frequency Change of TENS (a) increase efficacy (b) unchanged with use (c) decrease Patients using (a) regularly TENS (b) occasionally outdoors (c) never Patients using TENS with analgesic drugs Incidence of skin reactions (i.e., irritation of rash)

No. of patients

128

48 48

129

162

(a) (b) (c) (a) (b) (c)

72 29 27 28 15 5 35

: = h 5

10

ac 5

0 P

:: v” Z_LC_~_~I~-Y ::

N

0

9

% ,”

-I

VI

a

z

2

z

P-

m

g

VAS Score b Total relief

No relief

of pal” Fig. 1. Pain relief achieved

of pal” using TENS in 16X chronic

pain patients.

This may be due in part to a psychological biasing of patient response to visual analogue scales. All the patients who achieved no analgesic benefit from TENS (VAS = O-l) nevertheless expressed a desire to continue TENS treatment. In fact two-thirds of the patients who reported complete failure of TENS analgesia continued to use the stimulator on a daily basis! No significant differences (l-way analysis of variance, ANOVA) were found between the degree of pain relief achieved with TENS when patients were classified according to anatomical region of pain, aetiology of pain or diagnostic clusters. Although 75% (117) of patients reported daily stimulator use, only 52% (61) of these had used it on the

(%)

(56) (23) (21) (58) (31) (11) (73)

(a) 13 (IO) (b) 75 (58) (c) 41 (32) (a) 85 (53) (b) 31 (19) (c) 46 (28)

147

110(75)

143

45 (31)

Hours of use per week Fig. 2. TENS usage during

an average

week in 152 pain patients

Time

AN0V.A) were found between the mode clt ~?~~n~datv~n and the anatomical region of pain, aetroiog of pain or diagnostic clusters Thirty-two per cent of parrenta (41~ Itported a dccline in TENS efficacy since the time ,)f r. 0.05) was found between sensory threshold and pulse frequency. Males were found to have a significantly higher sensory threshold than females (mean + S.D.: male = 12.9 i 6.9: female = 9.2 _t 5.9: unpaired t test P c 0.01). No significant differences in sensory threshold (l-way ANOVA) were observed across regions. aetiology or diagnostic clusters Therapy level The intensity of stimulation, required to achieve the therapeutic analgesic effect, was corrected for individual variations in sensory threshold by simple subtraction: Therapy level (mA above sensory (mA) -- sensory threshold (mA)

0

2

(hours)

of post TENS analgesia

in 150 chronic

pain patients.

threshold)

7 absolute

therapy

levei

A positive correlation was found between sensory threshold (mA) and therapy levels (mA above sensory threshold; Pearson correlation coefficient r -= + 0.48, df = 86, P < 0.01). The mean + S.D. therapy level for all patients was found to be 9.1 rt: 9.1 mA above sensory threshold. No relationship (Pearson correlation coefficient r = + 0.2, df = 86, P > 0.05) was found between therapy level and pulse frequency. Males utilised a significantly higher therapy level (mean + SD.: male 11.4 + 11.1; female 7.0 + 6.4; unpaired t test P c 0.05). The intensity of TENS to achieve analgesia is shown in Fig. 5. Seventy per cent of patients (61) utilised therapy settings below 10 mA although a small number of patients exceeded current settings of 40 mA. No significant differences in therapy levels (l-way ANOVA) were

225

Intensity of TENS (mA above sensory threshold) Fig. 5. Intensity

found across anatomical tic clusters respectively.

region,

aetiology

of TENS to obtain

and diagnos-

Pain threshold measurements, taken at the site of TENS electrodes, were also corrected for sensory threshold in each individual: (mA above sensory threshold) - sensory threshold (mA)

in 87 chronic

pain patients.

Percentage therapy level

Therapy levels were recalculated as a percentage falling within a window between sensory threshold and pain threshold. Thus,

Pain threshold

Pain threshold threshold (~4)

analgesia

= absolute

%Therapy level = absolute therapy level (mA) - sensory threshold (ti2 pain threshold (mA) - sensory threshold (mA)

x loo

pain

Mean & S.D. pain threshold for all patients was found to be 18.20 rt 15.2 mA above sensory threshold. No relationship (Pearson correlation coefficient r = + 0.2, df = 86, P > 0.05) was found between pain threshold and pulse frequency. Pain threshold was highly correlated with therapy level (Pearson correlation coefficient r = +0.77, df= 86, P < 0.01). Males showed significantly higher pain thresholds (mean f SD.: males = 22.5 * 17.6; females = 14.4 + 11.6; unpaired t test P c 0.05). No significant differences in pain threshold (l-way ANOVA) were found across anatomical region, aetiology or diagnostic clusters.

No relationships (examined by plotting data, Pearson correlation coefficients and ANOVA) were found to exist between percentage therapy level and any of the other variables recorded. It should be noted that ail of these threshold and intensity measurements were taken at the TENS pulse frequency adopted by the patient. In an attempt to allow for the heterogeneity of pulse frequency across the patient group, the values of sensory threshold, therapy level, and pain threshold recordings were each multiplied by the value of the corresponding pulse frequency. These transformed values were subjected to statistical methods as mentioned above. No additional relationships were found to exist.

226

Freyuenc:v of .rtimulution (Hz)

TABLE

II

Mean i_ S.D. (mA)

The frequency of pulse delivery for burst (measured as pulse delivery within the ‘burst’ or ‘train’ of pulses) and continuous modes of stimulation were found to be highly correlated (Pearson correlation r = + 0.743, d’= 89. P < 0.01). Fig. 6 shows the pulse frequency used by patients when set to continuous mode stimulation. Over 75% (68) of patients utilised frequencies between 1 and 70 Hz with a small cluster of patients, 13% (12) choosing 111~ 140 Hz. Mean i S.D. frequency of pulse delivery was found to be 53 I 48 Hz (median = 32 Hz; mode = 24 Hz). As all stimulators used by the patients are constructed with a frequency control dial with logarithmic characteristics, it is difficult for patients to adjust the instrument to specific pulse frequencies above 40 Hz. Such frequencies lie on the steep part of the frequency output curve, thus small turns of the frequency control dial result in large changes in frequency [lo]. As a consequence, the distribution of pulse frequencies used by patients may directly follow the characteristics of the frequency dial. All patients utilising pulse frequencies between 111 ancl 140 Hz were using Tiger Pulse or Tiger Burst stimulators and had reached the maximum possible frequency available. No significant differences in pulse frequency were found across anatomical region, aetiology and diagnostic clusters. No correlations (Pearson correlation coefficient) were observed between frequency of stimulation and the present pain rating, as measured on a VAS (r = + 0.001. df= 89, P > 0.05), the degree of pain relief achieved using TENS (VAS; r = +0.132, df= 89, P > 0.05) or any of the personality variables measured by the EPQ.

Senwry threshold (n = 88) Therapy level * (fl = X7) Pain threshold * (n = 83) * mA above sensory

Low freq”e*cy (20 Hz)

11.6,

8.1

High freq”lXlcy (100 Hz)

x.7 t

Paired

I test

5.4

P i 0.001

11.1 k11.3

x.3 :t_ 9.9

P i 0.001

21.6 f 16.9

16.7 :t 16.3

P < 0.001

threshold

Follow-up Pulse frequency remained consistent over the followup period (Pearson correlation coefficient r = +0.66, df = 23, P < 0.01. Paired t tests showed no significant differences between visits, P = 0.93). Therapy settings declined between visit 1 and 2 (mean +_ S.D., mA above sensory threshold: visit 1 = 10.2 + 10.6; visit 2 = 6.1 +_ 6.2: paired t test P < 0.01) as did pain threshold although not to a significant level (mean _t S.D.; mA above sensory threshold: visit 1 = 18.7 f 16.9; visit 2 = 14.0 k 11.3; paired t test P = 0.07). Both therapy levels and pain thresholds were highly correlated between the two visits (Pearson correlation coefficients: therapy r = + 0.8, df= 23, P < 0.01; pain threshold r = + 0.7, df = 23, P < 0.01). No significant differences were observed for sensory threshold over the two visits (mean +_ S.D.; mA: visit 1 = 12.7 f 6.5; visit 2 = 12.2 + 8.9; paired t test P = 0.55; Pearson correlation coefficient r = + 0.82, df = 23. P < 0.01). Compurison of low frequency (fixed at 20 Hz) and high freyuencv stimulation (fixed at 100 Hz) Patients required significantly more current to attain sensory threshold, therapy level and pain threshold at the low frequency (20 Hz) of pulse delivery as shown in Table II. Males were found to have significantly higher sensory thresholds, therapy levels and pain thresholds for both high and low frequencies of stimulation.

Discussion

Fig. 6. Preferred

TENS pulse frequency

in 90 chronic

pain patients.

This is the first in-depth study of a large number of patients who successfully used TENS on a long-term basis for relief of a wide variety of chronic pain conditions. A striking finding was the remarkable lack of correlation between patient, stimulator and outcome variables. This result is consistent with the observations of Bates and Nathan [2] and Wolf et al. [23]. Although our patients were not compared with TENS non-responders, the present analysis reveals much information

221

of importance to clinicians involved in pain control by means of TENS. The following findings are of clinical relevance: (i) 47% of patients found that TENS reduced their pain by more than half; (ii) the onset of TENS analgesia occurred immediately in 30% of patients, within less than 0.5 h in 75% of patients and within 1 h in 95% of patients; (iii) post-TENS analgesia lasted less than 30 min in 51% of patients and more than 1 h in over 30% of patients; (iv) about 40% of patients used TENS for up to 30 h/week, whilst 30% used it for over 61 h/week; (v) 44% of patients benefitted from the availability of burst mode stimulation; (vi) 75% of patients utilised pulse frequencies between 1 and 70 Hz (mean = 53 Hz, median = 32 Hz, mode = 24 Hz) although a smaller sub-group preferred 111-140 Hz. Some of these points are discussed below. Questionnaires In a study of patients with chronic facial pain, Eriksson et al. [5] found that men attained greater benefit from TENS than women. By contrast, Johansson et al. [9] found that neither age nor sex had predictive value. The latter finding is confirmed in the present investigation in which there was no correlation between these factors and the degree of response to TENS across any of the chronic pain conditions.

rapid offset (within 30 min of stimulator ‘switch off’ for over 50% of patients) was noted by Hansson and Ekblom [7]. This rapid ‘on-off analgesic effect accords with a spinal gating mechanism as proposed by Melzack and Wall [15] where the nociceptive input carried by small afferent fibres (AS and C) is inhibited from onward transmission within the dorsal horn of the spinal cord when large afferent fibres (AP) are activated by TENS. Eriksson et al. [4] found the induction time of acupuncture-like TENS (using trains or ‘bursts’ of pulses presented at a low frequency between 1 and 4 Hz) to be 20-30 min, and slower than the induction time of conventional TENS which was 2-10 min. In the present study the time to onset with burst or continuous modes did not differ. This discrepancy may be due to the different intensities of burst stimulation used. Thus in Eriksson’s study stimulation sufficient to produce muscle contraction was employed whereas our group of patients used weaker intensities below that required to induce contraction. Post-TENS analgesia This is an important therapeutic bonus but in over 50% of our patients it lasted less than 30 min as observed previously by Andersson et al. [l]. However a third of our patients achieved post-TENS analgesia of over 1 h, possibly as a result of activating an endorphin release mechanism [4,20]. Over half of the patients preferred continuous mode TENS alone to treat their chronic pain and only a quarter used the burst facility alone. No significant differences were found between the duration of post-TENS analgesia obtained with burst or continuous stimulation. Many patients reported discomfort while using the burst mode possibly due to the occurrence of phasic muscle contractions.

Degree of analgesia Long-term results of peripheral conditioning stimulation as reported by Eriksson et al. [4] indicated that three quarters of chronic pain patients who obtain successful analgesia reached over 50% pain relief (measured on a VAS). Approximately half of the patients in the present study reported over 50% relief of pain during TENS use. Surprisingly 11 of the 23 patients who reported no TENS analgesia continued to use the stimulator on a daily basis, thus posing an interesting anomaly. Patients who report no pain relief on a VAS may not necessarily represent failures of TENS treatment because some of them still receive benefit evidenced by their statements that ‘TENS does not reduce my pain, but it distracts, or takes my mind off it.’ Such patients may produce idiosyncratic VAS scores and reveal a deficiency in such ratings. C,ur patients utilised a wide range of treatment regimes. Some used TENS for over 9 h daily proclaiming the success of the ‘wonderful black box’! However, despite the expe,rience of long-term stimulator use, some patients still showed anxiety on using TENS, possibly reducing efficacy as suggested by Nathan and Wall [16].

Long-term efficacy The declining response to TENS with time, often termed tolerance to TENS analgesia, has remained a fundamental problem for clinicians and patients alike. Although the mechanism of this tolerance is far from understood, it may be due to an adaptative change by the nervous system to regular repetitive stimuli produced by TENS [3,22]. In the present study 32% of patients reported a decline in TENS efficacy from the time the stimulator was issued, although for the majority (58%) TENS efficacy remained unchanged. Attempts to prevent or delay the occurrence of tolerance by the use of random electrical stimulation have led to the development of a new generation of stimulators such as the Codetron [IS] and the Likon. although these devices remain to be fully validated.

Onset of analgesia The rapid onset of TENS analgesia (within stimulator ‘switch on’ for 75% of patients)

Site of stimulation When patients were first issued with a stimulator, they were instructed to apply electrodes directly over, or

30 min of and also

immediately proximal, to the site of pain 121,231. Patients were also encouraged to experiment with electrode positions in order to find the optimal site for pain relief. Subsequently most patients had carried out an exhaustive trial of different electrode positions and were generally using sites originally suggested to them; none had stimulated remote body sites. Frequency of stimulution Andersson et al. [l] found that the degree of pain relief was related to the frequency of stimuiation so that. for example, 2 Hz was insufficient to produce analgesia. Linzer and Long [12] reported that 74% of patients used frequencies between 1 and 60 Hz, although it is generally stated that frequencies between 50 and 100 Hz are the most beneficial for patients [13,23]. Our patients favoured the frequency band l-70 Hz with an additional cluster at 111-140 Hz. An earlier study carried out in this Unit [lo] found that frequencies between 20 and 80 Hz produced the greatest analgesia measured as an elevation of cold-induced pain threshold in normal subjects. Unfortunately specific frequencies above 40 Hz are usually difficult to attain on commercially available stimulators because these lie on the steep part of the frequency output curve which has a logarithmic function so that a small turn of the control dial results in a big change in frequency [lo]. The distribution of pulse frequency used by patients in this study mimicked the characteristics of the frequency control dial. Different models of stimulators had different frequency output characteristics and these probably influenced the patients’ choice of stimulation frequency. Thus the cluster of patients using frequencies between 1 I1 and 140 Hz were all using Tiger Pulse and Tiger Burst stimulators whose maximum frequency setting lies in this range. Moreover. analysis of the questionnaire indicates that patients prefer to use faster frequencies of stimulation. This fact coupled with the lack of use of frequencies between 41 and 110 Hz suggests strongly that patients were not always able to attain the maximal preferred frequency to achieve the largest analgesic effect. It is clear that there is considerable scope for improved design of commercially available stimulators. There is good evidence that each patient prefers a particular pulse frequency to treat his or her pain condition. This is shown by the consistency of choice within the follow-up period of both frequency and pattern of stimulation and by the report from 73% of patients that they regularly reset pulse frequency before use. Further evidence is that patients were observed to return to their preferred frequency after completing the recording of sensory, therapy and pain thresholds at high and low frequency TENS (see Methods -- Frye analysis). This frequency preference may be attributed to achieving maximal analgesia (perhaps ,manifested in a more pleasurable sensation of stimulation). Since no correla-

tions were found hstuern pulw t‘reyuetx~ ;~ntf auazt~u: cal region of pain. xtiology i)C pain,