Physical Therapies in the Decongestive Treatment of Lymphedema

Study Design: the study was a multicenter, randomized controlled study conducted at three university hospitals in Spain. Potential subjects were identified at Rehabilitation Department where physicians were responsible for screening subjects for eligibility and enrolling into the study.

Patients that fulfilled the inclusion criteria and signed up the informed consent were stratified within 2 stratums, upper and lower limb lymphedema, and then were randomized to one of the three treatment groups.

The type of randomization was a total randomization with ratio 1:1:1 and the method used to generate the random allocation sequence was a computer-generated random numbers table. Adequate randomization includes calling a central office once eligibility has been determined.

The physicians that assessed the outcome and the statistic who analyzed de data were blinded to the subjects' treatment assignment.

The study was approved by the Research Ethics Committee of the University Hospital La Fe. It was conducted according to the ethical principles laid down in the declaration of Helsinki (52nd World Medical Association General Assembly, Edinburgh, Scotland, October 2000) and the rules of best clinical practice.

Treatment Regimens: Patients were randomized to one of the following treatment groups:

• Group A or control group: Manual Lymphatic Drainage (MLD) with Földi's technique by an expert therapist during 45 minutes; intermittent pneumatic compression (IPC) with multicompartmental pump between 50 to 80 mmHg during 30 minutes, followed by multilayer, multicomponent bandages (MB) until next day.
• Group B: Pneumatic massage with Lymphapress-Plus(TM) device that was lent during the duration of the study during 20 minutes, and was supposed to open the proximal lymphatics at the root of the limb; IPC (50 to 80 mmHg during 30 minutes), followed by MB until next day.
• Group C: IPC (50 to 80 mmHg during 30 minutes) followed by MB until next day.

All patients were planned to receive 20 sessions of treatment (5 per week in consecutive days during 4 consecutive weeks) and were trained in skin care and daily exercises.

After DLT, garments were prescribed and their adaptation was checked to warrant the maintenance of the results. The garments were flat knitted, custom made, generally class 2 (18-21 mmHg) for upper limb, and class 3 (36-46 mmHg) for lower limb.

Sample Size and Statistical analyses: This trial compares the efficacy of two treatment modalities without MLD (experimental group B and C) with a standard trimodal therapy (control group A). The expected effect size in percentage reduction in excess volume (PREV), primary end point, for the control group (MLD+IPC+MB) was defined from the review of literature and was 50 to 70% (mean 60%) with a standard deviation of 20 to 30 (mean 25). Our hypothesis was that there are no differences between experimental and standard treatment.

The required sample size for the study was 177 patients or 59 subjects per group to detect differences greater than or equal 15% in the percentage reduction in excess volume (PREV) between control and experimental groups, if exists; considering differences less than 15% as not clinically significant. It was based on an alpha error of 5% and a power of 90%. We estimated a drop-out rate of 10%. Therefore, 17 patients were added, resulting in a sample size of 194 subjects.

To compare baseline characteristics between three groups we use two-way ANOVA test for continuous variable and Pearson's chi2 for categorical data, to detect potential confounding factors. Outcome data were analyzed utilizing a two-way ANOVA test to assess differences between treatment groups with a 5% level of significance (p≤0.05) and two-tailed tests.

An exploratory subgroup analysis was made to identify factors associated with response (PREV). Factors for initial screening were identified by univariate linear regression model with a p value <0.1. In order to determine the independent predictive factors, a multivariate linear regression analysis was applied.