Participants have been recruited face-to-face at point of care during clinical visits from New York University (NYU) Cancer Center. To accomplish recruitment of 120 participants, we plan to use the successful procedures of recruiting and consenting participants used by the PI and the team in the preliminary studies [3,9,14,17,19]. Successful strategies include the use of Invitation Flyer that describes the study. This Invitation Flyer is posted on the bulletin boards or breast cancer support website at the cancer center, and is also available in the reception areas of the cancer center, examination rooms, and rooms holding support group meetings. In addition, health care providers such as nurses, oncologists, breast surgeons, and oncology radiologists at the center are willing to refer women meeting the inclusion criteria to the study by distributing the Invitation Flyer that describes the study to the potential participants.

After reading the Invitation Flyer, if a woman is interested in participating in the study, she would schedule a meeting with the research coordinator at that time or at other convenient time for them. During the meeting, the research coordinator will confirm her interest, determine if the woman is eligible for the study and the research coordinator will again explain the study in detail and provide enough time for the woman to ask questions. If the woman agrees to participate, she will sign the consent form.

Confidentiality will be maintained. Patients will be assigned a study ID specific for the study. Study data recording for the research will only use the study ID without the patients’ identifying information. A document file that has the patients’ study ID with the patients’ identifying information will be separately stored in locked files accessible only to the research coordinator, research nurse, or the PI. Electronic data will be stored in a password protected computer accessible only to the PI, the research nurse, and the research coordinator. Data analysis will be carried out and reported in the aggregate data so that individual identities are not revealed. Careful training and supervision of research coordinator and research nurse will insure study procedures are carried out in accordance with established protocols.

The Web- and mobile-based The-Optimal-Lymph-Flow [19,22] includes information about lymphedema, diagnosis and measurement of lymphedema, lymphatic system, risk of lymphedema, self-care, daily exercises, arm precautions, and ask experts. Participants in the intervention group will have access to the 8 Avatar videos that provide step-by-step instructions for The-Optimal-Lymph-Flow exercises to promote lymph flow and optimize shoulder and limb mobility. The platform also has a section entitled Arm Precautions, representing current patient education that emphasizes precautionary lifestyle behaviors, such as avoidance of repetitive limb movement, lifting weighted objects, needle punctures, blood draw, and the use of compression garments for air travel in the affected limb [20,21]. Figures 2-5 shows some screenshots of the The-Optimal-Lymph-Flow program. Table 1 presents the strategies, rationales, and actions for The-Optimal-Lymph-Flow program.

Patients assigned to The-Optimal-Lymph-Flow intervention group will have access to the Web- and mobile-based The-Optimal-Lymph-Flow platform and they will learn about theprogram and daily exercises during the first in-person research visit. Patients will have access to the website contents of Lymphedema, Diagnosis of Lymphedema, Lymphatic System, Self-care, Daily Exercises, and Ask Experts. Patients will also have access to the 8 avatar videos that provide step-by-step instructions of daily exercises to promote lymph flow and optimize shoulder and limb mobility. In addition, the patients will be introduced to an app, and have the choice to use either the Web-based program or the app for daily exercises. However, patients will not have access to the section Arm Precautions since the participants in the intervention group will receive comparable information as in the Arm Precautions section, but with particular emphasis on “what to do,” rather than “what to avoid.”

Patients assigned to the control Arm Precaution group will have access to the website section that emphasizes on precautionary lifestyle behaviors, such as avoidance of repetitive limb movement, lifting weighted objects, needle punctures, blood draw, and the use of compression garments for air travel in the affected limb [20,21]. Patients will have access to the following contents of the website: Lymphedema, Diagnosis of Lymphedema, Risk of Lymphedema, Lymphatic System, 3 avatar videos for Daily Exercises to promote limb mobility, and Arm Precautions. However, patients will not have access to The-Optimal-Lymph-Flow program, including Self-care, Daily Exercises to promote lymph flow, and 8 Avatar videos as well as Ask Experts that allows them to email to the researchers about their self-care. Patients will access to the Web- and mobile-based Arm Precaution program and learn about the program and daily exercises to promote limb mobility during the first in-person research visit.

From the previous one-arm clinical trial and usability and feasibility studies on the Web- and mobile-based The-Optimal-Lymph-Flow [19,23], we estimate that it would take about 45-60 minutes for patients to learn all the sections of the program and about 15 minutes to learn the The-Optimal-Lymph-Flow exercises for the intervention group through 8 avatar videos. It takes about 5 minutes to perform a set of The-Optimal-Lymph-Flow daily exercises each time. We encourage patients in the intervention group to perform at least twice a day or more times of the exercises during the study period. Participants in the control group will have access to three limb mobility exercise avatar videos and it takes less than 3 minutes to perform a set of limb mobility exercises each time. Patients in the control group will be instructed to perform limb mobility exercises at least twice a day or more times during the study period.

Data will be collected at baseline prior to intervention, and at week 12 post-intervention. Data collection at each time point will take approximately 30 minutes. Within one week of enrollment for the clinical trial, patients will have baseline assessment of pain and symptoms, limb volume difference, BMI, and quality of life. The follow-up assessment will occur at week 12 post intervention.

Two In-Person Research Visits: (1) prior to intervention there will be baseline assessment of pain and symptoms, limb volume difference, BMI, and quality of life; and (2) 12 weeks post-intervention assessment of pain and symptoms, limb volume difference, BMI, self-care behaviors, and quality of life.

Two Online Assessments: patients in the intervention and control group will receive an email that provides a link to assess pain and symptoms as well as quality of life at week 4 and week 8 post-intervention. Confidentiality of the patients will be protected for the online assessment since patients will use their study ID to access the online assessment.

Demographic and Medical Information: a structured tool is used to gather demographic and medical information and is verified through reviewing participants’ medical records [14,17,19]. The demographic and medical information will be considered as covariates, including expectation of the program, pain medications, age, surgeries, lymph nodes procedure, radiation, chemotherapy, time since surgery, time since lymphedema diagnosis, and hormonal therapy.

Primary and Secondary Outcome Measures: primary measure focuses on pain which is assessed during prior to the intervention and week 12 post-intervention in-person visit, as well as week 4 and 8 post-intervention online assessment. Secondary measures include symptoms, limb volume difference by infra-red perometer, BMI, quality of life related to pain. Limb volume difference by infra-red perometer and BMI are only measured prior to and week 12 post-intervention in-person visits. Symptoms and quality of life related to pain are assessed prior to and week 12 post-intervention in-person visit as well as week 4 and 8 post-intervention online assessment.

Pain and Symptoms Related to Lymphedema: the Lymphedema and Breast Cancer Symptom Experience Index is a valid and reliable self-report tool to assess pain, including aching, soreness, tenderness, as well as symptoms related to lymphedema (ie, arm swelling, breast swelling, chest wall swelling, heaviness, firmness, tightness, stiffness, burning, stabbing numbness, tenderness, stiffness, redness, blistering, and tingling (pins and needles) [14,17,19]. Each symptom can be treated as categorical variable by choosing a “Yes” or “No” to indicate the presence or absence of a given symptom. Each item can also be rated on a Likert-type scale from 0 (no presence of a given symptom) to 4 (greatest severity of a given symptom). Higher scores indicate more severe symptom presence. A response frame of last three months will be used for all participants to ensure the chronicity of symptom presence during the first in-person visit.

Limb Volume Difference by Infra-Red Perometer: perometry 350S will be performed on each arm as it is held horizontally. The perometer maps a 3-dimensional graph of the affected and non-affected extremities using numerous rectilinear light beams, and interfaces with a computer for data analysis and storage. A 3-dimensional limb image will be generated and limb volume will be calculated. This optoelectronic method has a standard deviation of 8.9 ml (arm), less than 0.5% of limb volume with repeated measuring [17,19] .

Quality of life Related to Pain: the Pain Impact Questionnaire (PIQ-6), a reliable and valid six question health survey, will be used to measure pain severity and the impact of pain on an individual's functional health and well-being. The PIQ-6 measures the severity of pain and its impact on work and leisure activities, as well as on emotional well-being within a variety of diseases and general populations. High PIQ-6 T scores indicate greater pain impact/worse health [21-23].

Height, Body Weight, and BMI: height will be measured to the nearest 0.1 cm with a portable stadiometer (Scale-Tronix 5002 Stand on Scale, Scale-Tronix Company, Carol Stream, IL, USA) without shoes [25]. An electrical device (InBody 520, Biospace Co, Ltd, Seoul, Korea) will be used to measure the participants’ body weight, BMI is calculated using the formula: weight (kg)/height (m²) [25].

Practice of Self-Care Behaviors: Risk Reduction Behavior Checklist is a structured self-report checklist that will be used to quantitatively and qualitatively assess patients’ practice of self-care behaviors at the study endpoint of 12-week after intervention [17,19]. The checklist include a list of self-care behaviors that promote lymph flow, eg, muscle-tightening deep breathing, muscle-tightening pumping, shoulder exercises, large muscle exercises, and having nutrition-balance and portion-appropriate diet, adequate hydration and sleep, as well as compression therapy for lymphedema.

The primary endpoint for the study is absence of pain or pain reduction reported by the participants at week 12 post-intervention.

We will enroll a total of 120 participants: 60 participants in The-Optimal-Lymph-Flow intervention and 60 participants in the Arm Precaution control group to account for a potential attrition of 20%, which has been observed in the prior studies in breast cancer survivors [10]. This will yield an adequate analytic sample size. Even with 20% attrition based on a 2 sample 2-sided t test with alpha=.05 and power of 90%, we can detect a difference of 0.7 standard deviations in the difference between the presence of pain in the intervention group compared to the control group at 8 weeks or at 12 weeks. The projected sample size will also provide sufficient statistical power for mixed regression models. For linear mixed models of continuous outcomes (eg, pain ratings), statistical power will exceed 80% to detect a medium effect, assuming a constant group effect, correlations of r=.5 between observations, alpha=.05, and compound symmetry of the covariance structure.

For binary outcomes, based on the three repeated observations with a conservative estimate for the assumed correlation of r=0.5 between observations, alpha=.05, sample size of n=50 per treatment arm will have power of 80% to detect odds ratios of the difference between groups ranging from 2.6 to 3.5 (small to medium effects).

Adequacy of sample size will be monitored during the analyses in three ways: (1) assessing sufficiency to estimate the number of model parameters identified by preliminary analyses; (2) examining model fit indices (eg, intra-class correlation coefficient to assess fit of random effects) to ensure data adequately support models generated, and (3) assessing the adequacy to accurately estimate model parameters by generating confidence intervals.

We will summarize graphically and numerically the distributions of pain, aching, soreness, tenderness, other symptoms related to lymph fluid accumulation, limb volume differences, BMI, and quality of life as well as covariates (such as medication for pain, age, or education and treatment variable and limb volume and self-care behaviors). The proportion of individuals experiencing pain, aching, soreness, tenderness will be compared between the intervention and control group over time, as will the mean severity ratings and associated quality of life scores. Mean comparisons of number of reported symptoms, limb volume, and BMI will be conducted between the intervention and control group as well as over time.

Bivariate relationships between the variables and treatment group will be assessed using point-biserial correlations and phi coefficients to identify potential covariates for inclusion into statistical models. Linear mixed effects models will be used to analyze continuous outcomes (eg, ratings pain, aching, soreness, tenderness, symptom ratings, and quality of life) and generalized linear mixed models will be used to analyze binary outcomes (presence of pain, aching, soreness, tenderness, and total scores of quality of life). These models will incorporate fixed effects for time, group, and any identified covariates. As indicated by preliminary models to estimate variance effects, models will include random effects for subject-specific slopes and intercepts. A statistically significant fixed effect for group by time interaction will indicate a treatment effect, with the direction of difference determined from mean values. Models will be compared for goodness of fit and modifications to link functions, distributional form, and correlation structure will be made as necessary.

No interim analysis will be conducted since the intervention is only 12-week long.

Data from participants who are missing >20% of any scale will be excluded from calculation of that scale, though remaining data meeting requirements for completeness will be retained. Analysis of missing data will first determine whether it can be assumed to be missing at random (MAR) or not missing at random (NMAR). Based on the results of this step, appropriate methods will be selected for addressing missing data (eg, Heckman Selection for NMAR, Multiple Imputation for data which are MAR).

Further, mixed effects regression models are robust in the presence of missing data. Unlike traditional repeated measures designs (eg, repeated measures ANOVA) which employ listwise deletion, excluding all records for individuals who miss a single observation, mixed effects models are capable of incorporating all completed observations for estimations. Although our previous studies suggest that we will have a limited number of missed observations, individuals need not have the same number of observations. This maximizes the statistical power of our analyses and reduces the likelihood of systematic bias in estimates.