Patients with knee pain due to OAk will be recruited through advertisements in local media. Temporal summation, before and after a heterotopic noxious conditioning stimulation, will be measured. In addition, pain on a numeric rating score, WOMAC subscores for pain and function and global perceived effect will be assessed. Patients will be randomly allocated to one of two treatment groups (tens, sham tens). Follow-up measurements will be scheduled after a period of 6 and 12 weeks.

Tens influences pain through the electrical stimulation of low-threshold A-beta cutaneous fibers. The responsiveness of central pain-signaling neurons of centrally sensitized OAk patients may be augmented to the input of these electrical stimuli. This would encompass an adverse therapy effect of tens. To increase treatment effectiveness it might be interesting to identify a subgroup of symptomatic OAk patients, i.e., non-sensitized patients, who are likely to benefit from burst tens.

Transcutaneous electrical nerve stimulation (tens) is a non-pharmacological, inexpensive and safe form of analgesia [7]. The pain modulating effect of tens is assigned to peripheral components which may be regulated by central mechanisms [8]. The inhibitory effect of tens is based on the 'Gate Control Theory' of pain perception as described by Melzack and Wall [9]. This theory suggests that stimulation of large (A-beta) afferent cutaneous fibers activate the inhibitory-interneurons in the dorsal horn of de medulla. This may weaken the transmission of nociceptive signals from small diameter A-delta and C-fibers. As OA is a dynamic process that involves phases of inflammation with possible increase of pain during these phases, tens may then be indicated as a facilitator for exercise. The use of tens to relieve knee pain in osteoarthritis of the knee (OAk) is recommended in various clinical guidelines as a conservative treatment to relieve knee pain [10-12]. However, Rutjes et al. [13] conclude in their meta-analysis that adequate evidence to support the use of any type of transcutaneous electrostimulation in patients with knee osteoarthritis is lacking.

Central sensitization is defined as "an augmentation of responsiveness of central pain-signaling neurons to input from low-threshold mechanoreceptors" [14]. Central sensitization embodies modified sensory processing in the brain and malfunctioning of descending pain-inhibitory mechanisms [15]. The importance of central sensitization as a potential underlying mechanism of pain in OAk has recently gained interest [16]. In the past decade suggestions were already made concerning the influence of sensitization of wide dynamic range (WDR) dorsal horn neurons when explaining an increase of the mechanical pain threshold after applying a blockage of A -beta nerve fibers [17]. Continuous and intense nociceptive input from the OA-damaged knee joint may encourage central sensitization [18,19] and is assumed to play an important role in OA [20]. Central sensitization in patients with OAk was detected by Arendt-Nielsen and colleagues [16] by enhanced temporal summation (TS) of pain and impaired diffuse noxious inhibitory control (DNIC). TS is defined as an increase in pain rating after repetitive stimulation at a constant stimulus intensity [21]. It is thought to be a psychophysiological correlate of wind-up, which is defined as the increase in response magnitude of second-order nociceptive neurons and higher structures [22] to repetitive noxious stimulation [23]. DNIC is a phenomenon in which pain from one part of the body inhibits pain elsewhere in the body [21]. It appears that multireceptive WDR neurons in the dorsal horn of the spinal cord and trigeminal nociceptive play a key role in DNIC inhibitory processes [24]. These neurons are important convergence sites for both excitatory and inhibitory influences arising from more than one type of tissue and can be activated by both innocuous and noxious stimuli [25]. It has been suggested that DNIC functions as a filter, which allows the system to focus on a painful stimulus in a background of basic somesthetic activity. The mechanism appears to involve the spinal cord as well as supraspinal regions [25].

A community sample of OAk patients with knee pain will be recruited.

Following baseline measurements, subjects will be randomly allocated to one of two treatment groups, ie., tens or sham tens. However, all patients will be told that two different kinds of tens are being tested. Randomization will be performed in blocks of four, stratified by sex and age. The allocation of this will be done by an independent researcher who will not interfere with other experimental procedures. We will use two boxes: one for each sex. In each box, subgroups will be made for five age categories: 50-59; 60-69; 70-79; 80-89; 90-99 years. Four numbered cards will be put in each age category. The numbering of the cards starts at one and ends at 40 and each number will correspond to a tens or a sham tens treatment. At the start of the study, each age category contains two tens and two sham tens treatments. Each time a new patient is included, a card will be taken out of the corresponding box and age category. When the four cards of one category are used, all of them will be put back in the box, so that a second round can be started. At each round, the numbered cards will be differently assigned to a tens/sham tens treatment by an independent researcher. The measurements will be carried out by a different researcher than the one that informs the patients concerning the treatment modalities. The information of the baseline measurements will remain concealed until the end of the study ensuring blinded assessment.

Sample size calculation is based on data from Law and Cheing [30]. They found a significant difference in the reduction of pain (2.9(4.7) points on a 0-10 VAS) when applying a tens current (pulse width: 200 μs, frequency: 100 Hz) on the knee of patients with OAk compared to a control group not receiving tens treatment. In this study the response within each subject group was normally distributed with standard deviation 4.7. If the true difference in the experimental and control means is 2.9, we will need to study 42 experimental subjects and 42 control subjects to be able to reject the null hypothesis that the population means of the experimental and control groups are equal with probability (power) 0.8. The Type I error probability associated with this test of this null hypothesis is 0.05. To take into account a 15% loss of data sample size is inflated up to 50 per group (n/0.85). Sample size is augmented by 10% per potential confounding factor (i.e. age and sex) [31]. The total sample size to be recruited will be 120 subjects.

The subjects that enter the tens -group will be asked to apply the tens therapy for at least 40 minutes continuously per day during 6 weeks. All patients in the intervention group will be informed that the intensity should be high enough so that an unpleasant but non-painfull sensation is acquired. The settings of the current will not change during the study. The patients that are assigned to the sham tens treatment will receive an inactive placebo tens therapy using a nonfunctional unit that appears to work but provides no stimulus. To blind the investigator as well as the patient, the sham tens device will deliver current for 30 seconds and then ramps down to no current. This approach has been studied and has been proven effective in blinding subjects and investigators to eliminate expectation bias [34]. All participants will be told that they may not feel anything after a while but that this does not mean that the machine is not working. All patients will be asked to come back after two and four weeks to assess their ability to precisely replicate the tens installation.