This study was approved by the National Health Group Domain Specific Review Board on 25 January 2016 (NHG DSRB Ref: 2015/01189).

A randomized controlled trial was carried out in a tertiary public hospital in Singapore. First-time mothers recruited from the postnatal wards were randomly allocated to one of the 3 intervention groups: web-based group (receiving web-based psychoeducational intervention plus routine care), home-based group (receiving home-based psychoeducational intervention plus routine care), or control group (receiving routine care). Mothers were excluded if they had identified physical or mental disorders before and during pregnancy, had complicated assisted delivery with fourth-degree perineal tear, gave birth to a stillborn child, or had a child with a congenital anomaly or medical complications [8]. The sample size of 204 was calculated based on the primary outcome of maternal parental self-efficacy, with 68 allocated in each group [12,16]. The recruitment period was from October 2016 to February 2017, and the last 6 months follow-up data were collected in August 2017. The study design and procedure have been previously reported in papers detailing the study protocol [12] and clinical effectiveness [16]. This study conducted a cost-effectiveness analysis from the health care provider’s perspectives alongside the randomized controlled trial.

The health outcomes used to measure the effectiveness of the interventions included maternal parental self-efficacy, social support, postnatal depression, and anxiety. The instruments used included the 17-item Perceived Maternal Parental Self-efficacy Scale (total score range 17-68, higher scores meaning better self-efficacy) [27,28], the 22-item modified Perinatal Infant Care Social Support Scale (total score range 22-88, higher scores indicating better social support) [28,29], the 10-item Edinburgh Postnatal Depression Scale (total score range 0-30, lower scores meaning less depressive symptoms) [30], and the Anxiety subscale of the Hospital Anxiety and Depression Scale (total score range 0-21, lower scores meaning less anxious) [31]. The higher scores of self-efficacy and social support and the lower scores of depression and anxiety indicated better outcomes. Detailed descriptions of the instruments and their validity and reliability can be found in papers previously published for this project [12,16].

The costs referred to in this study were viewed from a health care provider’s perspective, which included costs associated with the study interventions and costs related to the health service utilization to treat postnatal medical conditions in mothers and babies. Broadly, costs associated with the study interventions included those incurred during the development and implementation of interventions, ongoing delivery, and maintenance of the interventions. Costs related to the development and implementation of the web-based intervention include the purchasing of goods and services for hardware, software, educational materials, webpage design, maintenance, and work delivered by third-party service providers and were recorded from actual invoices. Costs of delivering the home-based intervention program were captured based on home visit human resources, material costs, and transportation fees.

Costs associated with health service utilization due to postnatal medical conditions were collected using the Questionnaire on Healthcare Service Utilization during 3 periods: within 1 month after childbirth, 1-3 months after childbirth, and 3-6 months after childbirth. All medical conditions that required health care services were captured and classified into 2 categories: infant-related conditions and maternal-related conditions. Service provider information was also recorded, which included government polyclinics, private clinics (including general practitioner clinics), hospital specialist outpatient clinics, hospital emergency departments, hospital inpatient admissions, over-the-counter pharmacies, and others. For every health service received by patients, data on full bills (ie, before any subsidies or insurance payments) were collected to estimate the health care provider’s full cost for each service. In Singapore, the public health care delivery system is not-for-profit. The entire bill size is the best estimate for the true cost incurred for delivering care. The entire bill size for certain health care services is not charged at a flat rate; it varies among patients depending on the duration of care delivery at each visit and the seniority of the personnel attending to the patient. For example, the cost of an outpatient visit is higher if the consultation time with a senior consultant is longer compared to a shorter consultation time with a junior consultant, although the unit cost for the medications and procedures is standardized.

The economic impact of national implementation of study interventions over the next 5 years was estimated by multiplying the average cost per the first-time mother by the estimated average number of first-time mothers in Singapore. The number of first-order babies born from 2012 to 2016 in Singapore ranged from 19,292 to 20,755, yielding an average of 20,000 first-order babies each year [33]. In our projection, we assumed that all 20,000 first-time mothers would be eligible for the psychoeducational program. To consider different acceptance levels of interventions, 3 program coverage levels, that is, 10%, 50%, and 100% for the eligible first-time mothers were adopted for economic impact analysis.

In total, 204 mothers were recruited, with 193 included in the economic evaluation (64 for web-based, 63 for home-based, and 66 for control group). The CONSORT diagram and details of the mothers’ characteristics have previously been published [16]. Briefly, the mean age of the mothers was around 30 years for all 3 groups, and half of them were of Chinese ethnicity. Over 80% of the first-time mothers obtained a bachelor’s degree or above (165/204) and were employed (175/204, 85%) during the study period. Nearly three-fourths of the mothers had monthly household incomes of more than SGD 5000 and the majority (178/204, 87%) were healthy and had no chronic diseases. Overall, around 58% (119/204) of the first-time mothers had normal deliveries. There were no statistically significant differences in all sociodemographic, clinical, and baseline data among mothers in the 3 groups. There were no significant differences in the baseline outcomes of self-efficacy, social support, postnatal depression, and anxiety among the 3 groups [16].

During the 6-month follow-up period, health services utilization data seeking various services concerning infant-related and maternal-related conditions showed similar occurrence rates among the 3 groups (Table 1). The number and percentage of zero–health service use mothers were as follows: web-based (28/64 44%), home-based (20/63, 32%), and control (30/66, 46%). The average health care cost per participant was SGD 339.90 (SD 493) for the web-based group, SGD 375.40 (SD 699.9) for the home-based group, and SGD 411.20 (SD 601) for the control group. Moreover, the web-based program involved an upfront cost of SGD 17,210 to develop both the web design and educational videos and an annual web maintenance fee of SGD 2900 per year. Similarly, the home-based program required a SGD 233.40 upfront one-time training fee for the payment of an experienced Registered Midwife to train the Registered Nurse who delivered the face-to-face postnatal psychoeducation; home visits cost SGD 77.42 for 1-hour session, which is inclusive of the manpower cost as well as the costs of printed materials and transportation. When accounting for all relevant costs, including 6 months’ health care utilization cost, the web-based program cost an average of SGD 390.80/participant, while home-based and control programs cost SGD 437.80/participant and SGD 411.20/participant, respectively.

The effectiveness of the web-based and home-based psychoeducational intervention programs has previously been reported [16]. Point estimates of the ICER for the scores of self-efficacy, social support, postnatal depression, and anxiety are reported in Table 2. In Table 2, the adjusted effectiveness of each outcome (self-efficacy, social support, depression, and anxiety) was estimated by adjusting for baseline values, age, number of days since the baseline, ethnicity, employment status, monthly household income, prenatal courses attendance, and skin-to-skin contact with the baby. The base-case analysis showed that the web-based group incurred the least cost and was the most effective among the 3 groups across all outcome measures. It saved around SGD 81 per participant (18% of the average adjusted cost) compared to the home-based program (adjusted cost: web-based group SGD 376.50 vs home-based group SGD 457.60) while showing better performance in all the outcomes, and SGD 41 per participant compared to the control group (adjusted cost: web-based group SGD 376.50 vs control group SGD 417.90) with better performance in all the outcomes. Hence, it was considered dominant across all the outcomes. When comparing the home-based group to the control group, the ICER ranged from SGD 8.60 per social support score improvement to SGD 90.20 per anxiety score (Anxiety subscale of the Hospital Anxiety and Depression Scale) reduction. When the validity of the programs was set for 3 years, the web-based program remained the dominant strategy to provide postnatal care (Table 2). The same pattern was observed when the overall program cost was varied at 0% and 50%. At 200% and 300% of the original overall program costs, the control program became the most cost-effective strategy, followed by the web-based and home-based programs.

A probabilistic sensitivity analysis was adopted to account for uncertainty from the trial. Cost-effectiveness planes comparing web-based versus home-based and home-based versus control were also generated (Figures 3 and 4). As shown in Figure 3, the ICER estimates for self-efficacy and social support were more in the right-side quadrants when comparing web-based versus home-based interventions, and the ICER estimates for maternal depression and anxiety were more in the left-side quadrants. Similar patterns were observed for the comparison of home-based versus control groups (Figure 4). The cost-effectiveness acceptability curves (Figure 5) showed the probability of cost-effectiveness for the 3 groups at various willingness-to-pay levels for the scores of self-efficacy, social support, postnatal depression, and anxiety. The pattern of the cost-effectiveness acceptability curves for the self-efficacy, social support, postnatal depression, and anxiety scores were similar, with the highest curve from the web-based program and the lowest curve from the control program. As the web-based program was the most effective and was associated with the lowest average total cost among the 3 programs, it was most likely to be cost-effective at all willingness-to-pay levels. However, its probability of cost-effectiveness showed a flat trend when the willingness-to-pay level reached around SGD 100 per self-efficacy or per social support (per score change on the scale) and SGD 400 per maternal depression or per maternal anxiety (per score change on the scale). The lowest curve for all 4 outcomes was for the control group, which indicated that the routine care was the least likely to be cost-effective than the other 2 intervention programs.

As shown in Table 3, during the 5-year projection period, the web-based intervention was the least costly program at all 3 coverage levels (ie, SGD 3,430,711 at 10% coverage, SGD 17,026,711 at 50%, and SGD 34,021,711 at 100%), and based on the 100% coverage, the reduced total cost reached nearly SGD 7.1 million (41,120,000 minus 34,021,711) and SGD 11.3 million (45,282,233 minus 34,021,711) compared to the control and home-based groups at the end of the fifth year, respectively. Although there was a relatively high upfront cost for the web-based program, it needed a very low maintenance fee for each year after the first year and provided almost unlimited concurrent access to the web-based contents for all target populations. On the contrary, the home-based program required intensive manpower to deliver face-to-face teaching for each visit, which was sensitive to the program’s coverage and comprised 17% of the total cost. Therefore, even with the lower health care cost over 5 years, the home-based program was the costliest in terms of the total cost incurred based on the base-case analysis results.

To the best of our knowledge, this is the first study that economically evaluated modes of delivering psychoeducation to first-time mothers. Data on economic outcome measures were less frequently reported [34]. Findings from this study indicated that the web-based psychoeducation program was the most cost-effective approach compared to the home-based and control programs. In the base-case analysis, the web-based psychoeducation program dominated the other 2 programs across all outcomes. It saved around SGD 81 per participant (18% of the average adjusted cost) compared to the home-based program and around SGD 41 per participant compared to the routine care alone while showing better performance over them. Our findings are consistent with the results from previous studies demonstrating that web-based or internet-based intervention is cost-effective for various populations when compared to routine care or home-based intervention [10,11,19,20]. In addition, our findings are in line with results from systematic reviews and meta-analyses of the effectiveness and cost-effectiveness of eHealth interventions in somatic diseases, in which most studies indicated that eHealth was effective and cost-effective or at least promising [34]. Another systematic review and meta-analysis reporting the effectiveness and cost-effectiveness of eHealth interventions for depression and anxiety in primary care demonstrated that eHealth was only cost-effective for depression, but there was no evidence for its effectiveness for anxiety [35]. The findings are partly inconsistent with ours, which demonstrated that the web-based program was more cost-effective than the home-based and control programs for self-efficacy, social support, depression, and anxiety. In our sensitivity analysis, we also noticed that the probability of being the best cost-effective was between 45% and 50% for all 4 reported outcomes (Figure 5). Hence, our findings provided evidence that the web-based approach may be a cost-effective way of delivering psychoeducational interventions but need further confirmation with large scale data (eg, real-world study) [36].

To evaluate the budget impact of adopting the programs, we used the average cost data and the estimated number of first-time mothers in Singapore to project the cumulative cost over 5 years [11]. Assuming a 10% to 100% coverage of the web-based psychoeducation program, we found that around SGD 1 million to SGD 11.2 million would be saved compared to the home-based program or around SGD 680,000 to SGD 7.1 million costs would be saved when compared to the control program at the fifth year. The tremendous projected savings for the web-based program make it promising and attractive to be implemented nationwide. In short, over the 5-year projection, the web-based program showed long-term cost-savings compared to the home-based program and the routine care. Although the upfront cost was high for the web-based program, the low maintenance fee made it superior to the other 2 groups.

This study has several strengths, which include its rigorous study design and the use of market data for personnel costs. The cost-effectiveness analysis was conducted alongside the 3-group repeated measures randomized controlled trial with a 6-month follow-up period. To evaluate the robustness of the results, we conducted both deterministic and probabilistic sensitivity analyses to check the uncertainty of our model. When the overall program cost was increased by 2-3 times (200% and 300% of the overall program costs, respectively), the control group cost the least among the 3 programs. However, the ICER of the web-based program over control program was rather low, that is, ranging from SGD 1.50 to SGD 6.50 per score improvement among all outcomes for 200% of the overall program cost and from SGD 9.60 to SGD 41.60 per score improvement among all outcomes for 300% of the overall program cost. The web-based program could be a preferred approach, given these relatively low ICERs. The cost-effectiveness acceptability curves were constructed to assess the sampling uncertainty from the trial based on 10,000 times bootstrapped results. The probabilities of the web-based program being cost-effective increased when the willingness-to-pay threshold increased and remained the highest among all 3 groups regardless of the willingness-to-pay thresholds. Despite the strengths, there were some limitations in this study. First, medical utilization data were collected at an individual level for all participants during the follow-up period and captured all possible costs incurred. However, the sample size was estimated based on the clinical effectiveness of the primary outcome of self-efficacy rather than the cost-effectiveness; thus, this might have contributed to uncertainty surrounding the results [37]. Meanwhile, there were 11 participants with missing cost data for the analysis, which might have affected the estimation of the cost for each group. To better address the uncertainty, we conducted a series of deterministic and probabilistic sensitivity analyses and found that the results were quite consistent and robust across the various analyses. In addition, this study used a randomized controlled trial with a 3-group study design, and the effectiveness of the interventions was evaluated up to 6 months after childbirth (about 5 months after the intervention), thus contributing to the robust study design [12,16] and ensuring that the findings from this study were representative and meaningful. Second, this analysis estimated costs from a health care perspective and not the estimated costs from a societal perspective. Therefore, it is recommended that future studies should conduct cost-effectiveness analysis of similar intervention from a social perspective. Third, the measurements of the health services utilization were based on self-reports at each follow-up data collection, which might lead to an underestimation of health resource use in comparison with daily recording in diaries, as people may forget to recall the services used [10]. However, in this study, the recall periods were kept short. The recall period for the costs of health resource used was only 3 months. Moreover, since we used a randomized design in this study, this underestimation was likely to be equally distributed among the 3 groups. Therefore, it is unlikely that the ICERS were affected in this study. Lastly, the calculated costs of the projection of nationwide implementation of interventions over 5 years were based on all first-time mothers, including those with complications. Since we excluded these mothers with severe complications in this study, our estimated costs might be conservative since the actual health care costs for those with complications might be higher. However, this would not affect the group comparison outcomes.

Costs, clinical considerations, and participant preferences are often used to determine the choice of a treatment or intervention and its delivery [38]. The findings from this study suggest that a web-based postnatal psychoeducational intervention for first-time mothers was the most cost-effective approach for all outcomes measured. Currently, such postnatal psychoeducational interventions are not routinely provided in Singapore and need to be seriously considered in postnatal care to support mothers, especially first-time mothers. In addition, future studies using diaries to record the cost of using health resources and estimating costs from both health care and societal perspectives are recommended.

Our findings suggest that compared to home-based psychoeducational intervention and routine care, the web-based approach was promisingly cost-effective to deliver the postnatal psychoeducational intervention to first-time mothers and could be adopted by hospitals as part of postnatal care support.