OBJECTIVE: To implement a chatbot on the Facebook® platform to establish connections with healthcare professionals who had subscribed to the chatbot, provide medical and pharmaceutical educational content, and collect data for online pharmacy research projects. Facebook® was chosen because it has billions of daily active users which offers a massive potential audience for research projects.
PRACTICE DESCRIPTION: The chatbot was successfully implemented on the Facebook® platform following three consecutive steps. Firstly, the ChatPion script was installed on the Pharmind website to establish the chatbot system. Secondly, the PharmindBot application was developed on Facebook®. Finally, the PharmindBot app was integrated with the chatbot system.
PRACTICE INNOVATION: The chatbot responds automatically to public comments and sends subscribers private responses using artificial intelligence. The chatbot collected quantitative and qualitative data with minimal costs.
EVALUATION METHODS: The chatbot's auto-reply function was tested using a post published on a specific page on Facebook®. Testers were asked to leave pre-defined keywords to test its functionality. The chatbot's ability to collect and save data was tested by asking testers to fill out an online survey within Facebook Messenger® for quantitative data and answer pre-defined questions for qualitative data.
RESULTS: The chatbot was tested on 1000 subscribers who interacted with it. Almost all testers (n = 990, 99%) obtained a successful private reply from the chatbot after sending a pre-defined keyword. Also, the chatbot replied privately to almost all public comments (n = 985, 98.5%) which helped to increase the organic reach and to establish a connection with the chatbot subscribers. No missing data was found when the chatbot was used to collect quantitative and qualitative data.
CONCLUSIONS: The chatbot reached thousands of healthcare professionals and provided them with automated responses. At a low cost, the chatbot was able to gather both qualitative and quantitative data without relying on Facebook® ads to reach the intended audience. The data collection was efficient and effective. Using chatbots by pharmacy and medical researchers will help do more feasible online studies using AI to advance healthcare research.
OBJECTIVE: To estimate the clinical and economic implications of community pharmacy-based herpes zoster (HZ) vaccination services with a hypothetical scenario of non-pharmacy-based vaccination in the State of Utah.
METHODS: A hybrid model of decision tree and Markov models was used to estimate lifetime cost and health outcomes. This open-cohort model was populated based on Utah population statistics and included a population of 50 years and above who were eligible for HZ vaccination between the years 2010 and 2020. Data were derived from the United States Bureau of Labor Statistics, the Utah Immunization Coverage Report, the CDC Behavioral Risk Factor Surveillance System, the CDC National Health Interview Survey, and existing literature. The analysis was performed from a societal perspective. A lifetime time horizon was used. The primary outcomes were the number of vaccination cases increased, and the number of shingles and postherpetic neuralgia (PHN) cases averted. Total costs and quality-adjusted life-years (QALYs) were also estimated.
RESULTS: Based on a cohort of 853,550 people eligible for HZ vaccination in Utah, an additional 11,576 individuals were vaccinated in the community pharmacy-based scenario compared to the non-pharmacy-based vaccination, resulting in 706 averted cases of shingles and 143 averted cases of PHN. Community pharmacy-based HZ vaccination was less costly (-$131,894) and gained more QALYs (52.2) compared to the non-pharmacy-based vaccination. A series of sensitivity analyses showed that the findings were robust.
CONCLUSIONS: Community pharmacy-based herpes zoster vaccination was less costly and gained more QALYs and was associated with improved other clinical outcomes in the State of Utah. This study might be used as a model for future evaluations of other community pharmacy-based vaccination programs in the United States.
RESULTS: A total of 221 community pharmacists participated in the current study (response rate was not calculated since opt in recruitment strategies were used). Remarkably, nearly half of the pharmacists (n= 107, 48.4%) indicated a willingness to incorporate the ChatGPT into their pharmacy practice. Nearly half of the pharmacists (n=105, 47.5%) demonstrated a high perceived benefit score for ChatGPT, while around 37% of pharmacists (n= 81) expressed a high concern score about ChatGPT. More than 70% of pharmacists believed that ChatGPT lacked the ability to utilize human judgment and make complicated ethical judgements in its responses (n= 168). Finally, logistics regression analysis showed that pharmacists who had previous experience in using ChatGPT were more willing to integrate ChatGPT in their pharmacy practice than those with no previous experience in using ChatGPT (OR= 2.312, p= 0.035).
CONCLUSION: While pharmacists show a willingness to incorporate ChatGPT into their practice, especially those with prior experience, there are significant concerns. These mainly revolve around the tool's ability to make human-like judgments and ethical decisions. These findings are crucial for the future development and integration of AI tools in pharmacy practice.
OBJECTIVE: This review evaluated observational studies to inform the prevalence of SGLT2-inhibitor-associated DKA compared with other antihyperglycemic agents.
METHODS: A systematic review was conducted in PubMed and EMBASE until 19 July 2022 (PROSPERO: CRD42022385425). We included published retrospective cohort active comparator/new user (ACNU) and prevalent new user studies assessing SGLT2-inhibitor-associated DKA prevalence in adult patients with type 2 diabetes mellitus (T2DM) against active comparators. We excluded studies which lacked 1:1 propensity score matching. The JBI Checklist for Cohort Studies guided the risk-of-bias assessments. Meta-analysis was conducted based on the inverse variance method in R software.
RESULTS: Sixteen studies with a sample of 2,956,100 non-unique patients met the inclusion criteria. Most studies were conducted in North America (n = 9) and adopted the ACNU design (n = 15). Meta-analysis of 14 studies identified 33% higher DKA risk associated with SGLT2 inhibitors (HR = 1.33, 95% CI: 1.14-1.55, p < 0.01). Meta-regression analysis identified the study location (p = 0.02), analysis principle (p < 0.001), exclusion of chronic comorbidities (p = 0.007), and canagliflozin (p = 0.04) as significant moderator variables.
CONCLUSIONS: Despite limitations related to heterogeneity, generalisability, and misclassification, the results of this study show that SGLT2 inhibitors increase the prevalence of DKA among adult T2DM patients in the real world. The findings supplement evidence from randomised controlled trials and call for continued vigilance.
OBJECTIVES: This review evaluated the effectiveness of strategies for enhancing ADR reporting by healthcare professionals (HCPs).
METHODS: This systematic review was conducted following the Cochrane and the PRISMA guidelines. Five international databases were searched from inception to December 2023 and updated search to September 2024. Randomized clinical controlled trials (RCTs) and non-RCTs on enhancing ADR reporting were included. The primary outcomes were the number of overall ADR and high-quality ADR reports. Study quality was assessed using the EPOC risk of bias (ROB), and ROBIN-I for RCT, and non-RCT. All data were evaluated using a random-effects model, and heterogeneity was assessed using I2 statistic and chi-squared tests.
RESULTS: From 1,672 studies, 13 studies (10 RCTs, and 3 non-RCTs) with 28,116 participants were included. Two of 10 RCTs had low ROB while the remaining were judged as unclear and moderate ROB. Most studies were in high-income countries, and the main strategy was educating HCPs through workshops. Meta-analysis showed significant increases in overall ADR reporting through educating HCPs with a rate ratio (RR) of 5.09 (95%CI: 3.36-7.71, I2=84.5%, low certainty), and in high-quality reporting with 1.31 (95%CI:1.09-1.58, I2=0.0%, moderate certainty). Subgroup analysis indicated that educating HCPs through face-to-face workshops combined with the Tawai app (RR:10.5, 95%CI:8.74-12.61), a face-to-face workshop alone (RR:6.69, 95%CI:5.43-8.25, I2=0.0%), and repeated telephone (RR:2.59, 95%CI:1.75-3.84, I2=8.8%) significantly increased the overall number of ADR reports with moderate certainty. Email or letter communications showed no significant effect.
CONCLUSION: Educating HCPs via interactive strategies like face-to-face workshops with or without a mobile app and repeated phone calls improved ADR reporting. However, long-term, high-quality studies are needed to confirm these findings before recommending widespread implementation in clinical practice, especially in LMICs.