Patients were managed by MediHK, a WeChat-supported platform designed by the research team. An engineer from Hunan Normal University’s College of Information Science and Engineering provided technology support for building MediHK. MediHK has been patented by the National Intellectual Property Administration, People’s Republic of China (ZL 2015 1 0648320.2). MediHK contains 2 opening screens: (1) the patient interface (Multimedia Appendix 1) and (2) the medical interface (Multimedia Appendix 2). The former is for patients, and the latter is for the members of the pain management team, which consisted of physicians and pharmacists. The medical interface was designed to manage pain-related problems and provide consultation to patients in a timely fashion. Both interfaces included 3 modules: a user login module for inserting basic user demographics into MediHK; an e-consultation module for communicating between patients and medical providers; and an introductory module for MediHK education, which offers a quick response code for new users (Multimedia Appendix 3).

We conducted 3 rounds of consensus-building using Delphi methods [12] to determine patient-reported outcome measures (PROMs) that could be integrated into MediHK. The pharmacists sent messages to patients, as shown on the far-right side of Multimedia Appendix 2, and the patients would receive a reminder, as shown on the far-right side of Multimedia Appendix 1. Patients could consult on any questions about pain, and the pain management team would receive real-time WeChat messages and respond as soon as possible. The acceptable response time was generally within 2 hours. When patients needed a change in drug/dosage or treatment of an adverse drug reaction (ADR), pharmacists first reviewed the patients’ historical records through MediHK and, then, made recommendations and reminded the physician. In China, pharmacists have no right to prescribe. If the physician had conflicting opinions, an agreement would be reached through offline contact; then, the physician could adjust drug-therapy regimens. All treatment recommendations from the pharmacists and physicians were based on the guidelines of the European Society for Medical Oncology Standard diagnosis and treatment of cancer pain, the National Comprehensive Cancer Network, and the European Association for Palliative Care.

When the patient’s expression was not clear, there was no guarantee that the same physician or pharmacist would communicate back. However, since the previously submitted forms and consultation questions were available through MediHK, the new physician or pharmacist would review the patient’s history in all aspects. Patient information was protected by encryption. Note that all screenshots of the app include translations that have been added for clarity for this paper.

The inclusion criteria of participants included: (1) age ≥18 years; (2) diagnosis of malignant tumors by a pathological or cytological method; (3) pain that met the cancer pain diagnostic criteria according to National Comprehensive Cancer Network Guidelines and was moderate to severe (NRS ≥4); (4) ability of patients or their families to read Chinese and use WeChat; (5) a normal verbal ability and performance status; and (6) agreement to participate in the study and sign the informed consent form.

Exclusion criteria of participants were: (1) nonmalignant pain; (2) hepatic dysfunction (alanine aminotransferase ≥2.5×upper limits of normal [ULN], aspartate aminotransferase ≥2.5×ULN, or total bilirubin ≥1.5×ULN) or renal dysfunction (serum creatinine ≥2.5×ULN); (3) participation in other clinical trials; and (4) hospitalization during the 4-week trial period.

The primary outcomes included pain intensity and pain interference in daily life. Secondary outcomes included PROMs, medication adherence, ADRs, and rehospitalization rates.

All data were analyzed using SPSS software (version 22.0; IBM Corp), and all charts were made by the graphing software GraphPad (version 8.0.2 (263); GraphPad Prism). For measurement data, the normality test adopted the Kolmogorov-Smirnov method. If normally distributed, the data were expressed as mean (SD), and the comparison between the 2 groups used 2 independent sample t tests. If not normally distributed, the data were expressed as the median (IQR), and the comparison between groups underwent a Mann-Whitney U test. Counting data were expressed as a frequency and percentage. A chi-square (c²) test or Fisher exact test was used for comparison between groups. We screened for factors affecting the pain intensity of outpatients with cancer pain by multivariate linear regression analysis (backward method, in=0.05, out=0.10). We defined P<.05 (test level=.05, two-tailed) as statistically significant.

A total of 100 patients joined and completed this study, with 51 (51%) in the intervention group and 49 (49%) in the control group. Demographic information (ie, gender, age, height, and weight) and clinical information (ie, diagnosis, pain type, and site of pain) of the 2 groups were not statistically different, nor was the intensity, pain interference, and adherence to pain medications at baseline (Ps>.05; Table 1).

Pain intensity in the intervention group was significantly reduced compared with the control group. The worst pain scores, least pain scores, and average pain scores in the 2 groups were statistically different, with median values of 4 (IQR 3-7) vs 7 (IQR 5-8; P=.001), 1 (IQR 0-2) vs 2 (IQR 1-3; P=.02), and 2 (IQR 2-4) vs 4 (IQR 3-5; P=.001), respectively, favoring the intervention group. The difference in the present pain score of the 2 groups was not statistically significant (P=.23). However, the score of the intervention group was numerically lower than that of the control group (Table 2).

The number of forms submitted by the intervention group patients was much higher than that of the control group (710 vs 95), with an average of 4.64 forms per person per day vs 0.06 forms per person per day, respectively. The most common forms submitted by the control group were the BPI (53/95, 56%), pain diary (17/85, 18%), and medication list (15/95, 16%; Table 3). Even though the control group patients did not receive reminders to fill out the forms, they still actively contacted the pain management team through MediHK due to uncontrollable pain intensity, interference in daily life, or severe ADRs.

The complete adherence rate in the intervention group increased from 43% (22/51) to 63% (32/51), while that of the control group increased from 33% (16/49) to 51% (25/49; χ²=12.864; P<.001; Figure 1).

The overall incidence of ADRs was 36% (36/100) across the 2 groups at baseline and increased to 56% (56/100) at week 4. ADR incidence in the intervention group was significantly higher than in the control group (χ²=8.990; P=.003). In addition, 3 cases of intestinal obstruction and 2 cases of delirium were observed in the intervention group. Table 4 shows the distribution of ADRs between groups.

The 2 groups had a similar rehospitalization rates within the 4-week trial. There was no significant difference between the 2 groups within 4 weeks (χ²=0.010; P=.92).

We introduced possible factors that could contribute to pain intensity for each pain item in a multivariate linear regression analysis. The physician-pharmacist intervention through MediHK was an independent influencing factor for the most severe pain (β=–1.413; P=.005; Multimedia Appendix 8) and average pain (β=–1.154; P=.003; Multimedia Appendix 9). Aside from medication adherence, the intervention was significantly related to the least pain（β=–.701; P=.02; Multimedia Appendix 10). No factors significantly influenced the present pain (gender β=1.078, P=.16; age β=.018, P=.26; height β=.063, P=.18; weight β=–.017, P=.46; adherence β=–.282, P=.40; intervention β=–.598, P=.17; Multimedia Appendix 11).

The self-management of cancer pain is full of challenges, especially, for ambulatory patients. Approximately 70% to 90% of cancer patients can relieve pain adequately when carefully following the treatment guidelines. Now more fully developed, digital health helps to achieve good pain management in daily practice for ambulatory patients with cancer pain, particularly, in remote areas of China [6,11]. Patients’ various demands in supporting self-management help encourage the development of a multimodal web application [16,17].

This study included a joint physician-pharmacist team that managed ambulatory patients with cancer pain through a WeChat-supported platform, MediHK, with promising results. Even if the control group did not receive a reminder to fill out the forms, patients in this group actively contacted the pain management team through MediHK to determine whether the medication plan needed to be adjusted due to either uncontrollable pain intensity, interference on daily life, or severe ADRs. The results revealed the patients’ need to contact the professional team via MediHK for better pain management. The patients in the intervention group reported more ADRs compared with control group patients, primarily, because there were more reports obtained from intervention group patients. More ADRs were not in conflict with improving pain. For example, the pain management team added new drugs for pain treatment, which may have caused some ADRs, but most of them were tolerated after a few days and monitored closely by the pharmacist.

Yang et al [18] developed an app named Pain Guard for better pain management of discharged patients. Its functions were similar to MediHK, such as self-evaluation, real-time medication consultation, and record-keeping. The differences were that, for MediHK, we combined the NRS, Face Pain Scale, and Verbal Rating Scale to assess pain intensity accurately, while Pain Guard had only the traditional scale, NRS. We designed the module to record more medication-related details from patients, including drug name, dose, frequency, initial stop time, pain relief after medication, and adverse reactions. In addition, the physician or pharmacist could send forms embedded in MediHK, such as the ADR or adherence assessment form, to patients according to their status. These functions were unavailable in the Pain Guard. Scriven et al [19] used the BPI to evaluate patients’ pain while participating in the multisite telehealth group model. They found positive changes on the interference scale at the individual level (14% of patients) but no change at the group level. Another study offered standardized education and telemonitored for pain improvement, and BPI results indicated that, at 1 week, there were improvements in both the worst pain (from 7.3 to 5.7; P<.01) and average pain (from 4.6 to 3.8; P<.01) [20]. However, the portion of average pain rated ≥4 did not improve significantly because of the short study period [20]. Compared with telehealth, MediHK was more capable of real-time feedback.

Furthermore, we received more positive results because of our 4-week observation time. One study evaluated the effectiveness of pain management of a mobile phone app. Results showed that the pain relief rate was significantly different between the trial and the control groups (median 50, IQR 45-63 and median 0, IQR 0-25) [18]. Similarly, Sun et al [21] found a significant difference in the average pain score through an intelligent pain management system (mean 2.5, SD 0.42 vs mean 2.8, SD 0.47; P<.01). These findings support our vision of making full use of prescient and promising internet platforms to manage cancer pain. In another study with an internet application consisting of a pain diary and a pain education and consultation module, the present pain intensity and the worst pain intensity of patients in the intervention group were significantly reduced within 6 weeks [6]. MediHK found similar positive results and included more details in the pain diary module, such as recording all patient medication information and pain self-assessments at all times. It is worth mentioning that these studies were based on the NRS for pain assessment. However, MediHK also embedded the BPI form to consider pain itself and the interference it caused. During the 4 weeks, the worst pain intensity, least pain intensity, and average pain intensity of intervention group patients significantly reduced, with an average decrease of 1-2 points. In terms of pain interference, the impact of pain on patients' general activity, mood, relationships with others, and interests reduced. However, the difference was not statistically significant when compared with the control group. Intervention group patients showed significant improvements in adherence after 4 weeks, resulting from active interventions, raised awareness of patients, and real-time monitoring of ADRs, which was more accessible in the home setting through MediHK.

Regrettably, we did not record the impact of education status and age in keeping medical records. Patients who had never received education may take longer to keep records. However, since the included patients or their families were all able to use WeChat proficiently, we believed that MediHK was feasible; for patients who were too old or unable to record, family members or caregivers would help to send the form. In total, we accounted for the universal applicability of MediHK when we developed it to ensure easy operation. The only complicated step was the switch between the interface. However, this was emphasized when training in the outpatient clinic.

Knowledge deficits, inadequate pain assessment, misconceptions of pain, complex environments, and infrequent communication with health care providers are barriers in pain management. A joint physician-pharmacist team operating through a digital health platform can improve it. The cancer patient pain assessment was complicated. It is necessary to select quantification tools and assess the cause, location, quality, and relieving or aggravating factors of the pain comprehensively. The time that physicians spend on each patient is limited, and it is difficult to provide long-term and continuous monitoring. The digital platform can better solve these problems. The platform trains patients to record their pain conditions in a more standardized and targeted manner. During clinical encounters, clinicians can spend more time addressing patients’ concerns in a meaningful way, rather than running through checklists of questions [22]. In addition, this platform promotes patient self-management. It allows patients to pay attention to the daily changes in pain and offers a digital tool to seek out the help of a professional team when suffering from an intractable pain or serious ADRs.

It is essential to consider the clinical workflow, security and liability, and the time-cost. We conducted a preliminary investigation and consideration in the early stage and carried out several rounds of related process optimizations and software improvements. In addition, when patients first visited the clinic, we would state that their physicians and pharmacists would provide the home services via the platform, and patients trusted this service. Finally, patients signed an exemption agreement and informed consent to ensure medical safety before using the platform. The satisfaction of patients and medical workers on such digital health platforms matters. One study designed a module in a mobile app to survey overall satisfaction, and the questionnaire was completed by participants at the end of the study [18]. Another study also assessed patient satisfaction about the convenience and helpfulness of using mobile systems, receiving technical software support, receiving consultant and training courses, and prompt responses for help; the results indicated that patients had a high level of satisfaction toward these kinds of digital tools [21]. Our preliminary idea was to evaluate satisfaction by embedding a questionnaire. For patients, this included assessing the pattern of the platform and the pain management team’s joint management, the content of medication education, the acceptability of response time, and the overall services. For the pain management team, this included evaluating the ease of operation of the platform, the acceptability of clinical workflow interference, and working-time costs. The questionnaire could contain an open-ended question, in which both patients and the pain management team are encouraged to provide suggestions regarding improvements to MediHK.

There were some strengths of this study. First, this study was a real-world randomized controlled trial conducted in a large ambulatory clinic of a tertiary hospital. All patients were clinically recruited and randomly assigned. The integration of PROMs has not been a feature of other eHealth and (web) application–related studies, allowing this digital health study to help advance this field. In addition, real‐time reporting can facilitate just‐in‐time interventions based on an individual's current circumstance or environment. This study achieved real-time communication between ambulatory patients with cancer pain and health care providers through MediHK, extending medical services to ambulatory patients as a pathway for the self-management in home settings.

The study had the several limitations. First, this study had abnormally high participation, which will not necessarily reflect what would happen when patients use the platform independently, because pharmacists would send daily notifications. Second, it was prospectively powered and conducted in a randomized manner, but inevitable confounding factors can exist in the real world. Multivariate linear regression can only explain a small part of the influence of different pain intensity types. Third, since the study was conducted in a single tertiary hospital, applying this approach in other clinical settings may require some individualization to meet specific needs. Fourth, the observation time of only 4 weeks limited the long-term application of the results. Fifth, this study lacked further assessment about buy-in from both patients and the pain management team.

The joint physician-pharmacist team operating through MediHK enhanced communication, optimized outcomes, and promoted self-management of patients in home settings. This study supports the feasibility of integrating the internet into patient self-management of cancer pain. In the future, it will be necessary to enlarge the sample size to further explore the long-term effects of this method on the self-management of ambulatory patients with cancer pain.