Study design
The study was conducted in two public outpatient clinics of two tertiary public hospitals (Wuxi Huishan District People’s Hospital and Affiliated Hospital of Jiangnan University) in Wuxi, Jiangsu Province from December 2 to December 20, 2023, during the peak mycoplasma infection season4,5. These outpatient clinics, serving as primary healthcare access points for residents, were selected due to the region's high prevalence of pediatric mycoplasma infections29. Site A's outpatient clinic is part of Wuxi Huishan District People's Hospital, a tertiary general hospital, while Site B's clinic belongs to the Affiliated Hospital of Jiangnan University, a tertiary Grade A general hospital. Both outpatient clinics provide comprehensive pediatric services, with Site B offering more specialized care and advanced diagnostic capabilities.
Before the start of the study, the research team trained healthcare providers (2 doctors from site A, 1 doctor and 2 nurses from site B) to clarify the research protocol. These sessions covered details about the pay-it-forward process, M. pneumoniae testing procedures, data collection, and results reporting. These trained providers then piloted recruitment, implementation, and monitoring. Following the pilot phase, each hospital independently implemented the study by following the study protocol. We reported our findings according to the CONSORT cluster-extension guidelines30.
Study participants
Participants in this study were children attending respiratory outpatient clinics in two selected public hospitals. They were recruited and eligibility screened by healthcare providers from guardians who accompanied their children to the clinic. The eligibility criteria were: 1) be no more than 14 years of age; 2) have symptoms consistent with Mycoplasma infection (e.g., fever, cough, sputum, runny nose.)31; and 3) have a legal guardian consent to participate in the study. Exclusion criteria included: 1) having cardiac, hepatic, or renal dysfunction; 2) receiving treatment with antibiotics or glucocorticoids two weeks before the visit; 3) concurrent presence of congenital disorders, and 4) previous participation in this pay-it-forward program.
Randomization and masking
This RCT employed a cluster design, with each cluster comprising ten children. The decision to use a cluster size of ten was based on discussions with stakeholders and considering the capacity of the participating healthcare facilities to manage and deliver the intervention effectively to each group. This cluster size was deemed appropriate to ensure robust data collection and analysis while maintaining feasibility in terms of resources and logistics, as demonstrated in our pilot and other studies 18.
We selected a cluster RCT for several reasons. The intervention is designed to operate at a group level, relying on the diffusion of behavior change through social networks and community interactions14,32. By randomizing clusters of individuals to either receive the pay-it-forward intervention or serve as controls, the study can capture the ripple effect and potential synergies that may arise from the intervention's implementation within the same groups. We conducted stratified randomization based on study sites, with 16 clusters randomly assigned to each site, randomly assigned to the pay-it-forward or control arms in a 1:1 ratio. The cluster in this study was defined as a group of ten eligible children who agreed to participate based on the order of their visit, which was used in our other previous pay-it-forward trial18. The allocation sequence (Supplementary file 1) was computer-generated and sealed in an envelope, which was opened by the healthcare providers upon recruitment of the first participant. The trial was conducted with blinded data analysts to prevent assessment bias, where data analysts were unaware of participant allocation to intervention or control groups.
Procedures
In the pay-it-forward arm, healthcare providers provided verbal information to participants' guardians about the importance of M. pneumoniae testing for child respiratory infections using a pre-designed, consistent message. Subsequently, healthcare providers explained the pay-it-forward program, including its concept, the opportunity for free M. pneumoniae testing, and the opportunity to donate money towards other children's M. pneumoniae testing. Participants' guardians were informed that the standard M. pneumoniae testing costs RMB 100 (~ $14 US dollars) and that previous participants' guardians had covered the cost of their child's test through donations. After the introduction session, participants' guardians were encouraged to contribute donations to support M. pneumoniae testing for other children, regardless of whether they opted for testing. All donations were voluntary, with no specified amount required. Since online payments are prevalent in Chinese hospitals, we offered QR codes for WeChat and Alipay to those who chose to donate. Donations were anonymous, and a summary of the donated amount was publicly accessible on the Social Entrepreneurship to Spur Health website ( http://cn.seshglobal.org?page_id=26420 ). Guardians of participants in the control arm received the same information regarding the significance of M. pneumoniae testing as those in the pay-it-forward arm. They were required to pay the standard RMB 100 ($14) fee for testing, following the hospital's regulations. Each participant received RMB 50 ($7) as compensation after finishing the questionnaire. Participants who were willing to be tested were directed to a collection area within the hospital. Throat swab samples from children were collected for M. pneumoniae testing. All samples were transported to the Wuxi Center for Disease Control and Prevention (CDC) laboratory for nucleic acid amplification testing (ZCON, Inner Mongolia, China). Participants were notified of their results via contact by CDC staff. Those who tested positive for M. pneumoniae were referred to their healthcare providers for appropriate treatment and follow-up care.
Data collection
After introducing the intervention before testing, all participants’ guardians were invited to complete a brief self-administered online questionnaire to collect information about the participants, including their sociodemographic characteristics and clinical symptoms. We also collected socio-demographic information from the participants' guardians through this questionnaire. At each study site, healthcare providers were required to complete a standardized daily administrative log, which included the number of participants who agreed to recruitment, the number of participants who had M. pneumoniae testing, the number of participants in the pay-it-forward arm who donated, and the amount of money donated by participants.
Outcomes
The primary outcome was the testing uptake of M. pneumoniae in two arms, as documented in the hospital's electronic medical records system. The secondary outcomes included the proportion of children in each arm who tested positive for M. pneumoniae, the proportion of M. pneumoniae testing uptake across subgroups, the proportion of participants in the pay-it-forward arm who donated, and the amount of donations.
Sample size calculation
Based on our previous studies, we consider a superiority margin of 20% to be clinically meaningful. Our prior research on the pay-it-forward intervention demonstrated that this approach was superior to the standard-of-care by a 20% margin17,18. To achieve an 80% power to detect a difference between the two arms, 160 participants are needed respectively in the pay-it-forward and standard-of-care arms. The test statistic used is the one-sided Z-test (not pooled). We set the interclass correlation at 0.02 and the significance level of the test was 0.025. The sample size was calculated by PASS 15. Detailed sample size calculation can be found in the study protocol.
Statistical analysis
Descriptive statistics were used to summarize the sociodemographic and behavioral characteristics of the study participants. Continuous variables were presented as means with standard deviations. Categorical variables were presented as frequencies and percentages. Chi-square tests (χ2 tests) were used to compare differences in categorical variables between the two groups, while Student's t-tests or Mann-Whitney U tests were used to compare differences in continuous variables, depending on the normality of the data distribution. A generalized estimating equation (GEE) model was used to assess the intervention's impact on Mycoplasma testing uptake, adjusting for potential confounding variables. The GEE model estimated the risk ratio, proportion difference, and their two-sides respective 95% confidence intervals for the primary outcome, comparing the pay-it-forward arm to the standard-of-care arm. Subgroup analyses were conducted to examine the intervention's impact stratified by caregiver’s education and income level, study sites, childbirth sex, and clinical symptoms at presentation with the GEE model estimating the risk ratio, proportion difference, and 95% confidence intervals within each subgroup, adjusting for potential confounders. These variables were chosen as they were considered as potential effect modifiers or unbalanced between the two study groups at baseline. All statistical analyses were performed using Stata (Version 17, StataCorp LLC, College Station, TX, USA). The superiority of the pay-it-forward approach was concluded if the lower bound of the 95% CI for the proportion difference was greater than the pre-specified superiority margin of 20%.
Economic evaluation
We assessed the costs of the standard-of-care and pay-it-forward arms using a micro costing approach, with costs reported in USD (1 USD = 7.14 Yuan, 2023). The cost of implementation was estimated by examining invoices and self-reports of wages and time from healthcare providers. The cost items were categorized as fixed (i.e., regardless of the number of tests completed) or variable (i.e., based on the number of tests completed). The financial cost was calculated by subtracting the donation or contribution in each group from the economic cost. The analysis was conducted from the healthcare provider’s perspective (Wuxi CDC). We reported the economic and financial cost per participant tested, the proportion of participants who donated, and the total donation amounts. We conducted the cost analysis using Excel 2019 (Microsoft, USA). Costing file is available in the supplementary file 2.