In the preliminary literature search, we retrieved 389,000 titles from Google Scholar and 156 from PubMed (Figure 1). These were imported into Mendeley 1.16.1 (by Elsevier) and duplicates removed. The titles were semiautomatically reviewed to remove papers about nonhuman or nonbirth related tools, which resulted in the exclusion of 388,082 entries. During the screening of abstracts for the remaining 1074 papers, we eliminated papers about tools that did not incorporate computers during labor monitoring use to leave 11 abstracts. Four abstracts identified in the manual literature search were added to the remaining 11 to get 15 abstracts used to identify papers for full review. At the full paper review stage, one paper (the Bacis program study by Horner in South Africa) was excluded because its subject tool was not designed for use during labor, which left 14 papers with all inclusion criteria for final analysis, as shown in Table 1.

As shown in Table 2, 4 out of 14 analyzed papers addressed labor monitoring tools still at the planning phase. We analyzed 7 papers about tools that had reached the implementation stage. The authors of the remaining 2 papers addressed the cost and impact on clinic workflow evaluation of one tool, QUALMAT eCDSS [21].

Five tools that were in the clinical trial or field testing [22-25] phase were classified under the implementation stage of development. Three tools had undergone formative assessment in form of user or cost evaluations [23,26,27]. We retrieved literature on a tool that was designed and tested in 2007, but we did not get publications on its advancement. One tool had a summative evaluation of its cost and impact on clinic workflow [28]. We did not get any tool with definitive summative evaluation, that is, pragmatic clinical outcomes (morbidity or mortality) studies.

The main parameters of focus in the analyzed papers were monitoring all labor parameters (7 papers) and fetal heart (3 papers). Four tools were designed to monitor one of three labor monitoring sections, whereas the rest could monitor multiple sections, as shown in Table 3. Authors addressing multiple labor monitoring sections were chiefly concerned with electronic versions of the partograph computing abilities. Therefore, their innovations potentially addressed the 15 parameters on a partograph. Those on fetal status focused on fetal heart monitoring with a cardiotocogram (CTG). The authors who addressed labor progress only specifically targeted cervical dilation and descent or station of a cephalic fetus.

All tools were intended for clinical diagnosis support based on algorithms. The least basic was the capability to take measurements or give alerts and reminders to maternity care providers [24,29,34]. However, 8 of 11 tools could also provide diagnosis and action suggestions to the user. The proposed SELMA tool could use machine learning models to predict diagnosis and outcomes for different contexts of use.

As shown in Table 4 , the tools were mostly usable on existing personal computer hardware, especially mobile gadgets such as phones, laptops, and desktop computers. Majority communicated through client-server networks, and 4 of 11 (36%) used stand-alone computers. Microsoft Windows was the most commonly used operating system in desktops, and Android was used in most mobile systems. One tool (PartoPen) uses custom-made software, whereas the application-based tools were developed in a Java environment. For half of the tools, the authors did not specify the software framework used in development. Of 11 tools, only 4 were reported as customizable to suit a user’s context of work.

The authors of the analyzed papers planned for adaptable tools. The intended context of tool use was stated as low-resource settings apart from the ANGEL shield. This computer program was made for and operated in a university hospital, but there were plans of rolling it out to nearby rural health centers. For 5 of 11 tools, it was reported that they could be customized to different contexts of use and even more functions added, where necessary.

In this review, out of over 380,000 papers, 14 qualified for the final analysis. They represented studies of 11 computerized tools capable of aiding health workers in maternity care. All labor parameters could be monitored by 7 of the 11 tools upon implementation. Only one tool had summative evaluations, which included cost and indicator studies. We did not find evidence of morbidity or mortality evaluation for any tool. Most tools used open source software that was also adaptable to common computers.

In this study, we chose to conduct a scoping review due to an ostensibly low volume of systematic evaluations and publications but with potentially more works on the subject. To this effect, many papers on pregnancy care applications were identified, although they were excluded from analysis for failure to meet other inclusion criteria. PubMed and Google Scholar formed the basis for the main search results due to their popularity among authors and a wide coverage of subjects. They were augmented with a manual search that indeed yielded more papers [22,23,32,33] included in the final analysis. The inclusion period of 2006 to 2016 was chosen to encompass the 5 years before and after the 2011 WHO call to evaluate all design and scale-up stages in eHealth [14], including e-labor monitoring tools.

Many papers were identified, which echo the findings of Kortteisto et al (2014), that is, mHealth is widespread [16]. However, similar to WHO and International Confederation of Midwives concerns in 2011 [13,14], only a handful addressed maternal labor and delivery monitoring. Moreover, as Hall et al found [17], the authors reported on tools that were in developmental stages, protocol preparation to field testing, without the definitive summative evaluation (pragmatic clinical outcomes studies) needed in health care research [35].

This state of affairs may be due to the human-intensive nature and high litigation potential of labor monitoring events that designing and testing a reliable computerized labor tool calls for more effort than is needed for an average medical condition. Another factor that could deter innovators is the difficulty in definitive summative evaluation of the tools in light of many confounders of labor outcomes. A similar situation in 2011 could have led the WHO to call for better research and evaluation of mHealth [14] such as labor monitoring tools. One should also be cognizant of the diverse labor monitoring contexts both within and without health facilities and communities, which were highlighted in the 2016 Lancet maternal health series [36,37]. However, the trend of research seems to be in a positive direction. From the works analyzed in this paper, the majority of the papers are authored after the WHO call to action, and all the documented evaluations were published after the call.

After the data collection for this review, another call to action on improving quality of maternity and newborn care was sounded through the Lancet maternal health series [37]. Furthermore, the potential of mHealth to help out is anticipated, with innovators urged not to be stifled by the fear of liability and litigation but to provide evidence-based tools for woman-centered care [18,38]. This series reinforced the findings of this review and stressed the need for further research for context-specific mobile childbirth monitoring tools.

Regarding the focus of the tools, it was obvious that the three main sections—fetal, labor progress, and maternal states—of labor monitoring received unequal attention. Monitoring the fetal condition, especially the fetal heart, was most researched, perhaps due to the discovery of cardiotocogram (CTG) that is efficacious in detecting fetal distress. CTG is sonoelectric and improving it through computerization was easier than inventing tools as was necessary for the labor progress. On the other hand, cervical dilation and maternal conditions are subjectively dependent on provider skills, and sociocultural or religious norms. Hence, the diversity of these conditions could hamper the design, testing, and development of widely acceptable tools to monitor labor progress and maternal conditions. With the recognition of the diversity of contexts of use but limited resources, it is prudent that generic tools are designed and developed or adapted for specific contexts [39].

The operating system platforms were generic in about half of the tools perhaps to cut development costs. This is also good for the diverse hardware that is increasingly portable. On the other hand, the programming software base was mostly undisclosed. This could be due to the early stages of development. Moreover, the authors were not yet committed to a specific program. However, the generic Java development environment was used in most specified cases. This would suggest that even the rest are more likely to use it in tool development.

The sources of data for this review were not exhaustive of all literature (eg, papers not written in English), and as such, we could have omitted some papers on the computerized labor monitoring tools. This is a drawback, but the main sources are wide enough for an adequate sample and similar reviewers [17,20,40] increasingly adopt this approach. We also started the search with broad terms and narrowed them down as we saved different papers for subsequent analyses.

In conclusion, the scope and volume of research on computerized maternal labor monitoring tools is likely to be narrow and small. Fetal heart monitoring seems to dominate over other labor management parameters. Most tools are designed to use affordable computing platforms, but there is hardly any summative evaluation of the available tools. This may imply a slow response to the call for developing and evaluating computerized labor monitoring tools that could reduce labor-related disease. Further research, including clinical outcomes studies and publication of results, is needed on computerized tools for use in comprehensive childbirth monitoring.