This study was a single-center randomized control trial (trial registration: ChiCTR1900021920) with assessments conducted at baseline and post intervention (3 months). Participants randomized to the intervention group received anticoagulation management through the Alfalfa App (“Internet+pharmacy care”), while the control group received offline management at a pharmacist-led anticoagulation clinic. Both groups were followed up for 3 months.

The primary outcome indicator, time in therapeutic range (TTR), was used to calculate the required sample size. Our previous research suggested that TTR for the web-based and offline groups was 73% and 60%, respectively [7]. With a power of 80% and a significance level of .05, 1:1 ration of test and control groups, a minimum of 201 patients per group is required, calculated using PASS 27 software. Assuming 10% dropout rate, each group needed at least 223 patients, for a total of 446 patients.

Participants were recruited from Fuwai Yunnan Cardiovascular Hospital. Patients diagnosed with valvular heart disease, who underwent valve surgery at the hospital received warfarin education from clinical pharmacists before discharge. After training, a validated questionnaire [16] was used to assess the patient’s understanding of warfarin management. If patients missed a question, they would be retrained and answered until all the questions were answered correctly. These patients were eligible for the study if they had been on warfarin treatment for at least 3 months, received warfarin management through either the Alfalfa app or pharmacist-led outpatient clinic visits, agreed to regular follow-up, had no serious bleeding or thrombotic events in the 3 months before warfarin treatment. The exclusion criteria included pregnancy, switching to another anticoagulant during the study period, and having 3 or fewer INR records during the follow-up period. After obtaining informed consent and completing the baseline questionnaire, participants were randomized into either the web-based group or offline group.

Participants were randomized using a Microsoft Excel–generated randomization list in a 1:1 ratio. After participants completed baseline assessments, a research team member (unblinded) assigned participants to a group based on the randomization list. Due to the nature of the intervention, the study was open labeled and blinding was not feasible.

After randomization, participants in the web-based group were instructed through WeChat (Tencent) to download and install the Alfalfa app on their smartphones. Detailed instructions on app usage were provided. The Alfalfa app (Alfalfa Health Management) was used for remote warfarin management in the web-based group; a detailed description of the app has been published earlier [14,15]. Patients selected a pharmacist within the app who would manage their warfarin dosage. The chosen pharmacist reviewed INR results uploaded by the patient and provided recommendations on dosage adjustments and the timing of subsequent INR tests. The coagulation management team consist of 4 cardiologists and clinical pharmacists specializing in anticoagulation from Yunnan Cardiovascular Hospital. A senior clinical pharmacist responded to patient queries within 30 minutes of receiving uploads.

Participants in the offline group were instructed to visit the pharmacist’s office for periodic INR testing. Warfarin dosage adjustments were supervised by the same senior clinical pharmacist who managed the web-based group.

All participants adjusted their warfarin dosage as needed and underwent periodic INR testing throughout the study. Participants in the web-based group could test INR at nearby hospitals and receive adjustment suggestions from clinical pharmacists through the app. Patients in the offline group were required to visit the hospital’s anticoagulation clinic. Warfarin dose adjustment protocols were standardized by the coagulation management team. Monitoring was conducted weekly during the first 2 weeks of therapy and then every 2-4 weeks. The target INR range was 1.7‐2.5 for patients with valve replacement.

All participants received free pharmaceutical care and advice during the study period. Offline group participants were exempted from consultation fees in the anticoagulation pharmacy clinic during the study period. After the study, offline group participants were recommended to use the Alfalfa app. Registration and use of the Alfalfa app remained free for participants during and after the study. And the app also provided a free “anticoagulation circle” for patients to access educational resources of anticoagulation treatment and communicate with one another.

The primary outcome was the percentage of TTR, calculated using the standard linear interpolation method [17]. TTR is a widely recognized measure of anticoagulant control quality, with TTR ≥60% indicating high-quality anticoagulation and TTR <60% indicating low-quality anticoagulation [18,19].

Data were analyzed using Excel 2010 and SPSS (version 17.0, IBM Corp) statistical software. Different statistical methods were applied depending on the variable type results. An independent 2-sample, 2-tailed t test was used for statistical analysis of continuous variables, such as age, TTR, and costs. A chi-square test or Fisher exact test was used for categorical variables, such as sex, conditions, diseases, target INR, efficacy outcomes, adverse events, and INR results. Continuous variables are presented as mean (SD), while categorical variables are reported as frequency (percentage). Statistical significance was set at P<.05.

This study complied with the Declaration of Helsinki and was approved by the Ethics Committee of Fuwai Yunnan Cardiovascular Hospital, Kunming, Yunnan, China (Ethics No. 2020-010-01). It was also registered with the Chinese Clinical Trial Registry (ChiCTR1900021920). All participants provided written informed consent before the study and signed an electronic informed consent form again upon registering for the Alfalfa App. Participants were allowed to withdraw from the study at any time during the research process. All medical records and test reports of the participants were stored in the hospital where they received care, and personal privacy information was strictly protected. No additional compensation was provided to the patients for participating in this study; however, some free services were offered during the research period (see the Measures and Incentives subsection). The datasets generated or analyzed during the study are available from the corresponding author upon reasonable request.

Between January 2020 and December 2021, a total of 457 patients from Fuwai Yunnan Cardiovascular Hospital were recruited for eligibility assessment, and 450 were enrolled in the study. Of these, 225 participants were randomized to the web-based group and 225 to the offline group. Participants were excluded if they completed registration but did not submit INR records (n=5), did not visit the pharmacist clinic (n=8), provided incorrect medication information (n=7), were pregnant (n=2), switched to another anticoagulant (n=3), or had fewer than 3 follow-up visits or INR records (n=20). This resulted in a total of 405 participants included in the final analysis (Figure 1). There were no missing data among these participants, and baseline characteristics were comparable across study groups (Table 1).

Table 2 shows that the mean TTR was significantly higher in the web-based group compared with the offline group (66.46% vs 46.65%, P<.001). In addition, the percentage of INR values within the therapeutic range was higher in the web-based group compared with the offline group (124/204, 60.78% vs 67/201, 33.33%; P<.001; Table 2).

After 1 week of distributing and reminding participants to complete the survey links, a total of 86 voluntarily completed questionnaires were collected. Of these, 59.3% (51/86) had used the Alfalfa app for over a year, and 77.9% (67/86) preferred to stay in contact with their doctor through the app. The vast majority (83/86, 96.5%) found the app convenient, noting the ability to receive timely responses from doctors without needing to visit the hospital and to view all their report records within the app. Approximately 80% (69/86) of the patients expressed positive experiences with the doctor’s responses (82/86, 95.3%), the completeness of records (75/86, 87.2%), and the free services (69/86, 80.2%).

All participants were satisfied with warfarin dose adjustment (100%, 86/86), timely responses from doctors (86/86, 100%), the report submission process (78/86, 90.7%), and the history records functions in the app (85/86, 98.8%). Nearly half of the respondents (42/86, 48.8%) indicated they would be willing to pay a fee ranging from ¥3 (US $0.4704) to ¥10 (US $1.568) for the app’s services. A large majority (77/86, 89.5%) expressed a preference for services provided by experts with senior professional titles. A few participants (10/86, 11.6%) suggested improvements, such as simplifying the app’s operation, adding a feature for video interactions with doctors (3/86, 3.49%), and establishing a more comprehensive medical history record function (1/86, 1.16%; Multimedia Appendix 2).

The main goal of this study was to evaluate the efficacy, safety, and economic impact of using the Alfalfa app, a smartphone app designed for dose consultation and adjustment in patients undergoing long-term anticoagulation therapy with warfarin. TTR was selected as the primary outcome, with a TTR of 60% or higher generally considered indicative of high-quality anticoagulant therapy [25]. In our study, TTR was significantly higher in the Alfalfa app group compared with the offline group (66.46% vs 46.65%, P<.001). While a study by Lee et al [26] reported contrary findings, our results align with the majority of telemedicine-based anticoagulation studies, which show improved TTR outcomes compared with traditional management methods [27-32]. Previous cohort studies in Eastern China also supported the efficacy of the Alfalfa App, reporting significantly higher TTRs in their intervention groups (61% and 79.35%, respectively) [15,33]. Our findings further confirmed the effectiveness of Alfalfa App in improving the quality of anticoagulation management.

Our research also demonstrated that participants in the web-based group had a significantly higher frequency of INR monitoring compared with those in the offline group (8.14 vs 4.47, P<.001). This indicates that the Alfalfa App facilitated better interaction between patients and health care providers, enabling participants to easily upload their INR records and receive timely feedback from clinical pharmacists. In addition, the app’s supervision and tracking features likely contributed to the regular monitoring observed. These findings are consistent with 2 other retrospective cohort studies, which also reported higher rates of therapeutic INR values in the web-based group [15,33]. However, compared with the other 2 retrospective studies, there was no consistent pattern in the distribution of extreme subtherapeutic, subtherapeutic, supratherapeutic, and extreme supratherapeutic INR values between the groups. The result indicates that the web-based management model helps improve the percentage of INR values within the therapeutic range (indirectly equivalent to TTR), but when the INR values fall outside the therapeutic range, they do not necessarily tend to be too high or too low. This suggests that there may be no direct association between INR value distribution and anticoagulation management methods, indicating a need for further exploration with larger sample sizes.

Our research also revealed the higher incidence of minor bleeding events in the Alfalfa app group, consistent with the results reported in a similar cohort study [15]. According to a meta-analysis of 12 randomized controlled trials, telemedicine combined with portable coagulometers significantly improved TTR and reduced the incidence of thrombotic events in cardiovascular patients [34]. However, several studies reported no statistically significant difference (P>.05) in bleeding or thrombotic events between the eHealth tools group and the conventional management groups [28,30,35]. eHealth tools can be broadly categorized into 4 types, namely computer-based support systems, electronic health records, telemedicine platforms, and mobile apps. Among these, mobile apps offer the greatest convenience, allowing patients to communicate with doctors or pharmacists at any time [36]. This may explain the higher detection rate of minor bleeding events in the Alfalfa app group, as patients were more likely to report symptoms to clinical pharmacists through the app in time, especially during the period of COVID-19 pandemic, when patients were less willing to go out for medical treatment. This made minor bleeding in the web-based group of the study easier to detect. The pandemic also limited our ability to fully assess clinical event outcomes due to the reduced sample size and limited participant enrollment.

Our findings also suggest that the Alfalfa App may reduce the cost per visit for anticoagulation management and save participants time in return visits. The convenience of using the app led to higher adherence to follow-up visits in the web-based group, resulting in a greater number of return visits per capita during the follow-up period. Although we did not calculate the incremental cost-effectiveness ratio due to a lack of authoritative data, the cost-effectiveness value suggests a positive economic outlook for the Alfalfa App. It should be noted that patients in both the web-based and offline groups were exempt from registration fees. If web-based services were to be charged, combined with the increased number of follow-up visits, this could potentially offset the cost savings from not visiting the hospital, and the overall expenses might even be higher [37], and more research is needed to explore the full economic impact of the app.

Two studies revealed that most patients were satisfied with the use of eHealth tools, which also improved their quality of life [38,39]. Consistent with these findings, our survey results of the experience and satisfaction with pharmaceutical care and Alfalfa App also showed that the vast majority of patients were highly satisfied with the dose adjustment, timely responses, and free service provided. The timeliness and convenience of the Alfalfa App were particularly appreciated. However, a small number of participants complained the app difficult to use. During the follow-up, 5 participants discontinued to submit their results, resulting in withdrawals. This suggests that both the app and anticoagulant management services have areas for improvement. With the advancements of machine learning, the integration of deep learning, reinforcement learning, and ensemble learning in warfarin management after heart valve surgery is becoming more prevalent [40]. Internet+pharmacy care through app has the potential to enhance efficiency, reduce time and economic costs, and improve the quality of medical care, especially by reducing the disparities among primary care providers.

Although our study demonstrates promising results regarding the efficacy of internet+pharmacy care through Alfalfa Appin improving anticoagulant quality among patients following heart valve surgery, several limitations must be acknowledged. First, because the study’s enrollment period overlapped with the COVID-19 lockdown, patients reduced their hospital visits. The significant decline in hospital patient flow from 2020 to 2022 confirms this trend. As a result, the study’s sample size was limited, preventing us from conducting subgroup analyses. Second, this was a single-center study conducted in southwestern China, which may limit the generalizability of the findings to other populations. Further prospective, randomized, multicenter studies are needed to confirm these results.

We estimated a sample size of 446, assuming a low dropout rate of 10%. By the end of the study, 405 participants were included in the analysis, and the actual dropout rate was within expectations. With strong effect sizes, we maintained adequate statistical power. However, due to limitations with the Alfalfa App, we were unable to send private information to participants, which prevented us from distributing the satisfaction questionnaire individually. As a result, only 86 questionnaires were collected. In the future, we aim to address this limitation to enhance the completeness of research data.

In conclusion, the “Internet+Pharmacy Care” model through the Alfalfa App can improve the quality of warfarin anticoagulation in patients after heart valve surgery. Compared with traditional outpatient service, web-based management does not increase the risk of major bleeding, embolism, or hospitalization events, and is both economical and cost-effective. Looking forward, we anticipate that machine learning and artificial intelligence will bring new opportunities for innovation in anticoagulation management. Future studies with larger sample sizes, multicenter designs, and long-term follow-up are needed to further validate the role of these technologies.