This exploratory pilot study used a pretest-posttest design with a nonequivalent control group. The study followed the TREND (Transparent Reporting of Evaluations With Nonrandomized Designs) checklist [18].

This study was conducted at an outpatient clinic in a 1200-bed university hospital in Bangkok. In 2025, a total of 256 individuals with osteoporosis received care at this hospital, where osteoporosis services are provided at the orthopedic outpatient department every Wednesday. Medical appointments are scheduled every 4 to 6 weeks.

During the study period, the comparison group continued to receive usual care. Nurses provided health education focusing on medication, diet, and exercise. This education was delivered orally, without the use of written materials or video clips for exercise demonstration. Individuals could confirm or reschedule their appointments through the hospital application.

The primary outcome was self-care, which was measured using the Self-Care of Osteoporosis Scale (SCOS) [7], a self-reported instrument.

Although family members participated in the intervention sessions, outcome measures were collected only from participants with osteoporosis. Therefore, the individual participant was the unit of analysis.

Data were analyzed using SPSS (version 29.0; IBM Corp). Normality was assessed using the Kolmogorov-Smirnov test, and all variables were normally distributed; therefore, parametric tests were applied. Baseline homogeneity between groups was examined using the chi-square test, Fisher exact test, and independent-sample t test as appropriate. Differences in overall self-care following the intervention were analyzed using analysis of covariance (ANCOVA), whereas differences in the 3 domains of self-care were examined using multivariate ANCOVA (MANCOVA). Statistical significance was set at P<.05. To facilitate transparency and reproducibility, statistical outputs from SPSS, including baseline comparisons and ANCOVA and MANCOVA results, are provided in Multimedia Appendix 2.

Prior to the main analyses, the assumptions underlying ANCOVA and MANCOVA were evaluated. Normality of residuals was assessed using the Shapiro-Wilk and Kolmogorov-Smirnov tests. The Shapiro-Wilk test indicated a slight deviation from normality (W=0.946; P=.03), whereas the Kolmogorov-Smirnov test indicated no significant departure from normality (P=.07). Although the Shapiro-Wilk test suggested a deviation from normality, ANCOVA and MANCOVA are considered robust to moderate violations of normality when group sizes are equal (n=23 per group); therefore, parametric analyses were retained. Linearity between the covariate (baseline self-care score) and the postintervention outcome was confirmed (F_1,43=113.914; P<.001). The assumption of homogeneity of regression slopes was evaluated by including a group × baseline self-care mean score interaction term in the ANCOVA model and was satisfied as the interaction was not statistically significant (F_1,42=3.626; P=.06). The Levene test indicated that the assumption of homogeneity of variance was satisfied for the overall self-care score (F_1,44=2.989; P=.09).

For MANCOVA, the Bartlett test of sphericity confirmed significant intercorrelations among the 3 self-care domains, supporting the use of a multivariate approach (χ²₃=47.3; P<.001). Pearson correlation coefficients among the 3 postintervention domain scores ranged from 0.739 to 0.861 (P<.001 in all cases), indicating no evidence of multicollinearity (r<0.90). Homogeneity of variance-covariance matrices, assessed using the Box M test, was violated (M=21.231; F₆=3.276; P=.003). Therefore, the Pillai trace was used to evaluate multivariate group differences because it is robust to violations of the assumption of homogeneity of covariance matrices. However, the baseline mean score for the diet and physical activity domain differed significantly between the groups (P=.04), whereas the baseline mean scores for the medication adherence and safety performance domains did not. To account for this baseline difference, the corresponding baseline SCOS domain mean scores were included as covariates in the MANCOVA model. Therefore, the postintervention group comparisons were based on adjusted mean scores after controlling for the corresponding baseline SCOS domain mean scores, minimizing the potential influence of baseline imbalance on the interpretation of the between-group differences.

Ethics approval was obtained from Chulalongkorn University (022/68) and the Faculty of Medicine of Ramathibodi Hospital (MURA2025/477). All participants provided written informed consent prior to participation. To ensure confidentiality, all data were stored in password-protected files accessible only to the research team. Participants in the intervention group received a stipend of US $8 to offset internet-related expenses associated with participation in the intervention.

A total of 46 participants were included in the study, with 23 (50%) in the intervention group and 23 (50%) in the comparison group (Figure 1). No participants dropped out of the program. The test assessing the homogeneity of participant characteristics and dependent variable (self-care) showed no significant differences between the 2 groups, indicating that they were comparable at baseline (Table 2).

ANCOVA was conducted to compare postintervention self-care mean scores between the intervention and control groups while controlling for baseline self-care mean scores. The results showed that the intervention group had significantly higher postintervention self-care mean scores than the control group (adjusted mean difference 0.97, 95% CI 0.88-1.06; F_1,43=459.29; P<.001; partial η²=0.914; Table 3).

MANCOVA was conducted to examine differences in the 3 domains of self-care between the intervention and comparison groups after controlling for baseline domain mean scores (medication adherence, diet and physical activity, and safety performance). The overall multivariate group difference was statistically significant, as indicated by the Pillai trace (V=0.939; F_3,39=198.45; P<.001; partial η²=0.939), suggesting significant differences between groups for all 3 domains after controlling for baseline domain mean scores. Compared with the comparison group, the intervention group demonstrated significantly higher postintervention mean scores for diet and physical activity (adjusted mean difference 1.02, 95% CI 0.89-1.15; F_1,41=245.04; P<.001; partial η²=0.857), safety performance (adjusted mean difference 1.04, 95% CI 0.88-1.20; F_1,41=171.20; P<.001; partial η²=0.807), and medication adherence (adjusted mean difference 0.81, 95% CI 0.68-0.95; F_1,41=144.54; P<.001; partial η²=0.779; Table 3).

The findings indicate that the multicomponent supportive-educative mHealth intervention was associated with higher postintervention self-care mean scores among individuals with osteoporosis. Higher adjusted mean scores were observed across all domains of self-care (medication adherence, diet and physical activity, and safety performance), with the largest between-group difference observed for diet and physical activity.

The observed between-group difference may reflect the combined contribution of several intervention components. These components included the provision of clear and reliable disease-related information, weekly nurse-participant interactions, family support, sustained attention, and digital reinforcement. Disease-related knowledge may support self-care by enhancing individuals’ understanding of their condition and increasing their confidence in managing it. This is consistent with the supportive-educative nursing system by Orem et al [13], which emphasizes teaching and guidance to strengthen self-care agency. In addition, the weekly nurse-participant interactions provided opportunities for participants to discuss concerns, receive feedback, and clarify misunderstandings, whereas family involvement and digital reinforcement may have strengthened self-care behaviors and sustained participants’ engagement throughout the intervention.

Family support may also have played a role in these observed differences. When family members learn alongside individuals with osteoporosis, this facilitates shared review and reinforcement of the content and supports engagement in self-care. This reflects the emphasis by Orem et al [13] on providing an environment that supports personal development and self-care. Although evidence on family support intervention for osteoporosis management is limited, findings from exploratory studies suggest that family involvement may support self-care [17]. When one person misunderstands or forgets information, others can provide reminders or exchange ideas, thereby supporting sustained self-care practices. In many countries, including Thailand, extended family living and close family ties are common.

Although higher postintervention self-care mean scores were observed in the intervention group than in the control group, both groups showed increases in self-care mean scores over time. Therefore, the observed between-group differences should be interpreted cautiously as factors other than the intervention may also have influenced the findings. Participants in the control group received usual care, and repeated completion of the self-care assessments may have increased awareness of self-care behaviors. To reduce potential confounding, participants were matched on sex, age, and back pain score, factors previously associated with self-care [6,9]. In addition, baseline self-care mean scores were controlled for in the ANCOVA and MANCOVA analyses. However, the results should be interpreted as indicating an association between the multicomponent supportive-educative mHealth intervention and higher postintervention self-care mean scores rather than as definitive evidence of a causal effect.

The multicomponent supportive-educative intervention delivered partly through Facebook provided accessible and flexible learning. Video-based content allowed for repeated viewing, whereas remote interaction with nurses enabled timely guidance and reinforcement. This continuous connection supported engagement beyond clinical settings and facilitated shared learning with family members. However, this finding contrasts with those of previous studies of app-based interventions that found no significant differences on physical activity, calcium intake, or vitamin D intake, possibly because of the lack of a personalized and interactive approach [16].

Participants’ experience provides insights into how the intervention influenced self-care. Participants found dietary guidance easy to understand and reported actively implementing fall prevention strategies; in contrast, exercise was less frequently adopted. In this study, both groups reported rarely engaging in exercise before the intervention. After the intervention, however, the intervention group reported sometimes engaging in exercise, whereas the control group continued to report rarely engaging in exercise. Although exercise can increase bone mineral density, enhance muscle strength, and improve body balance, thereby reducing the risk of falls, adherence remains challenging. Previous studies have reported that exercise adherence among individuals with osteoporosis varies from 59% to 100% even when expert coaching or accessible facilities are provided [20]. These findings highlight the need for additional strategies to promote exercise adherence.

This study has implications for practice. The observed between-group differences suggest that the multicomponent supportive-educative intervention delivered partly through Facebook may warrant evaluation in randomized, attention-controlled studies before clinical implementation. Nurses may consider using digital platforms to extend care beyond clinical settings and engage family members as active partners in care while recognizing that the effectiveness of this approach requires confirmation in more rigorous studies.

This study has several limitations. As an exploratory pilot study with nonrandom allocation, the findings should be interpreted with caution because factors other than the intervention, including the unequal intensity of support between groups, may have been associated with the observed between-group differences. The substantial effect sizes should also be interpreted cautiously as the relatively small sample size and reliance on self-reported measures may have influenced the magnitude of the estimated associations. In addition, recruitment from a single hospital may limit the generalizability of the findings. The medication adherence domain demonstrated low internal consistency at baseline (Cronbach α=0.56), which may have reduced the precision of the estimated associations for this domain. Self-care was assessed using self-reported measures, which may be subject to response bias. Participants and researchers were not blinded to group assignment, and contamination between groups may have occurred because participants were recruited from the same setting. Finally, the short follow-up period did not allow for evaluation of the long-term sustainability of the observed improvements. Future studies using randomized controlled designs, larger and more diverse samples, objective measures of self-care, and longer follow-up periods are needed to confirm these findings.

In conclusion, the multicomponent supportive-educative mHealth intervention was associated with higher self-care mean scores across multiple domains among individuals with osteoporosis. The observed between-group differences may reflect the integration of educational content, family involvement, ongoing support, and digital delivery. However, they should be interpreted with caution given the exploratory pilot study and relatively small sample size. Further rigorous randomized controlled trials are needed to confirm the efficacy of this intervention.