RETRACTED ARTICLE: Improving usability and pregnancy rates of a fertility monitor by an additional mobile application: results of a retrospective efficacy study of Daysy and DaysyView app

The menstrual cycle is one of the characteristic physiological processes of the female body and it is a central indicator of overall health in women of reproductive age. Continuous fluctuations of hormones result in commensurable physiological changes throughout the menstrual cycle. In the last decade, specific biosensor-embedded devices have been developed to assist women in monitoring, measuring and representing these aspects of their body. For such devices, the typical-use related pregnancy rate is still low but was significantly worse than the method-related pregnancy rate. This implies that usability and understanding of a method plays a major role in a fertility monitoring device and its safe effective use. It is reported, that trough the additional use of a mobile application the interest and motivation of a patient’s health behavior increases significantly. The contraceptive effectiveness of the fertility monitor (Daysy) has already been demonstrated in an independent trial. The result of the method related Pearl-Index calculation obtained in the present study (0,6) differs only a little from what is reported by Freundl and colleges (0,7). However, if the focus is on the typical-use related Pearl-Index, it has significantly improved from 3,8 to 1,3. Independently, the perfect-use efficacy (0,8) of Daysy was calculated in this study.

We conclude, that it is possible through the present technology of Daysy and the additional, optional use of DaysyView to improve usability and enhance the typical-, method- and perfect -use pregnancy rates.

The menstrual cycle is one of the characteristic physiological processes of the female body and it is a central indicator of overall health in women of reproductive age. Continuous fluctuations of hormones result in commensurable physiological changes throughout the menstrual cycle [1]. These include fluctuation in urine luteinizing hormone (LH), cervical mucus and body basal temperature (BBT). Women have been engaging in monitoring these physiological changes as signs of their fertility for many years. In the past, they have charted different signs of their menstrual cycle (temperature rise, cervical mucus changes) to determine the onset as well as the end of the fertile phase by pencil and paper (Symptothermal Method) [2]. In the last decade, specific biosensor-embedded devices have been developed to assist women in monitoring, measuring and representing these aspects of their body.

Most of these devices use a combination of BBT measurement and sophisticated statistical methods in addition to a comprehensive on-board database to identify the fertile and infertile phase of the menstrual cycle [3]. Other devices use the correlation between pulse rate and the menstrual phases to determine the individual fertile window [4]. The medical device Daysy (Valley Electronics AG, Zurich, Switzerland) is an electronic device that also exploits the described relationship between the menstrual cycle and fluctuations in body temperature by measuring and recording the BBT as an aid in ovulation prediction for planning and preventing pregnancy by identifying the fertile and infertile phase of the menstrual cycle. Since the advent of fertility monitors, the reliability and safety of such devices has been tested in different clinical trials [2, 3, 5, 6]. For example, in their retrospective clinical trial, Freundl, et al., concluded that the fertility monitors Babycomp and Ladycomp achieved a method-related Pearl-Index (PI) of 0.7 and a typical-use related PI of 3.8 over 12 months, placing in a similar safety range as the natural family planning method (NFP) [5].

Since Daysy is based on the fertility algorithm of Babycomp and Ladycomp from Valley Electronics GmbH, it was claimed that Daysy has a similar PI to these products that are ultimately bounded by the fertility awareness-based method (FABM) itself.

For some contraceptive methods, such as sterilization or copper intrauterine devices (IUD), the inherent efficacy is extremely high and proper that extreme low pregnancy rates are found in all studies (Table 1).

For the remaining methods, the typical-use related pregnancy rate is still low but was significantly worse than the perfect-use related pregnancy rate (Table 1). This implies that usability and understanding of a method plays a major role in a fertility monitoring device and its safe effective use. It is reported that through the additional use of a mobile application (app), the interest and motivation of a patient’s health behavior increases significantly [7]. One reason is that people have the tendency to interact with, or check, their mobile devices regularly; this repeated reviewing is reinforced by immediate visible information [8].

Tracking of menstrual cycles via app has been announced to be a common form of self-monitoring to either avoid or achieve pregnancy. In the United States about 80% of 18–49 year olds own a smart phone and approximately 28% are using mobile healthcare apps. Of 90.088 healthcare apps in the Apple iTunes store, 7% (6300) are for women’s health and pregnancy (Institute for Healthcare Informatics). Mobile fertility apps gain support, but the majority are lacking clinical evidence [9]. A current study reported by Setton and colleagues, concluded that apps used to predict the fertile window and dates of ovulation are generally inaccurate [10]. One reason is that there is a wide variation in each menstrual cycle. Even in women with a “perfect” 28-day cycle, the fertile window varies from cycle to cycle [11]. However, until the present time, no study has been reported considering the contraceptive effectiveness of a fertility monitor (Daysy) optionally connected with an App (DaysyView). This combination is interesting because as described above, it is shown in various studies that the use of apps is increasing patients´ focus on their disease or their health behavior. Mobile displays were effective in encouraging users to maintain activity level, and reminder notifications aimed at goal achievement were desirable features [12‐16].

In this study, we presented that through the additional use of an App and thereby improved usability of the medical device, it is possible to enhance the typical-use as well as the method-related pregnancy rates.

Within the study 6278 contacted women, 1969 (31%) followed the invitation and 798 (13%) completed the survey. The total number of recorded cycles was 4738. The average age of participants was 29 years (Fig. 1a), whereby the fertility monitor was mainly used to avoid pregnancy (74.68% see Fig. 1b). Because the size and weight of the participants were queried, it was possible to determine the body mass index (BMI; kg/m [2]), which was 23.02 on average (Fig. 1c). Women over 25 years had a significantly elevated BMI compared to women between 20 and 25 years (Fig. 1c) in this study.

From 798 participating women, 524 (64%) indicated an additional contraceptive use (Fig. 2a). Out of the 493 respondents using additional precaution methods, 73% (358 out of 493) used the method during the fertile (red) phase, 22% (109 out of 493) used the method during the fertile and the infertile (green) phase of the menstrual cycle and 6% (27 out of 493) stated to use the additional method inconsistently (Fig. 2b). The analysis of the data showed that the male condom (93%) was the most common additional contraceptive (Fig. 2c). Among the 493 women using an additional contraceptive method, 110 (22%) preferred several forms of contraception.

From a total number of 798 women using the fertility monitor for family planning, contraception or both, a total of 4750 cycles were identified. Through the internal database, all cycle data were double checked for their correctness. In five cases, the correctness (due to the lack of the serial number of the device) of the data could not be confirmed, these participants were excluded from the study.

668 respondents (2674 cycles in total) declared they had been using the fertility monitor for < 13 cycles (Fig. 1d). Their data was not taken into account when calculating the PI. Furthermore, 125 respondents (2076 cycles in total) reported they had been using the fertility monitor for > 13 cycles (Fig.1d). Among these 125 women, 2 women indicated that they had been unintentionally pregnant during the use of the device. Therefore, the total pregnancy- or typical-use related pregnancy rate is 2 × 1300 / 2076, which equals a PI of 1.252.

Counting only the pregnancies which occurred as a result of unprotected intercourse during the infertile (green) phase, we found 1 pregnancy, giving a method-related pregnancy rate of 0.626 according to the PI. To further calculate the perfect-use related pregnancy rate we calculate all cycles (1725 in total) in which the user was sexually active but stated to have had no unprotected intercourse on red (fertile) days for > 13 cycles. Therefore, the perfect-use pregnancy rate is 1 × 1300 / 1725, which equals a PI of 0.753.

The life-table shows for each month what the probability is that a woman becomes pregnant, as well as the typical-use related pregnancies by cycle (Table 2). The overall pregnancy-rates and its 95% confidence interval (CI) were calculated according to the Kaplan-Meier approach [19, 20]. The analysis shows that after 13 cycles of exposure, the typical-use related probability of an unintended pregnancy was 2.707% (Fig. 3 blue line). Focusing on women who claimed to have always had protected intercourse (independent of the fertility status), the probability of an unintended pregnancy decreases (n/s) to 1.92% after 13 cycles (Fig. 3 black line). The same value increases significantly to 10.82% probability if a woman is considered to have had unprotected intercourse on red (fertile) as well as on green (infertile) days (Fig. 3 imperfect use). If a woman had unprotected intercourse exclusively on green days (perfect-use related pregnancy rate), the probability of an unintended pregnancy was 2.19% (Fig. 3 perfect-use).

The average cycle length of all participants was 28.9 (± 3.52 SD) days. A closer analysis of the different age groups shows a significant difference in the cycle lengths (Fig. 4a). Thus, the length of the cycle decreases with increasing duration. The same trend is evident when women were asked for irregular cycles. When a fluctuation of 3–4 days is considered to be normal, 195 respondents (24.24%) reported to have irregular cycles. When women under 20 years are focused, the percentage of irregular cycles increases (41.18%). Because of the small number of participants under 20 (n = 17), the result is not significant (Fig. 4b). We found that when focusing all participants (n = 798) that 94.7% have a menstruation between 2 and 7 days. The result differs somewhat when only woman over 40 (n = 22) are considered. In this fraction of women, 10% (n = 2) have indicated that their menstruation lasts less than two days (Fig. 4c).

In the third section of the survey, women were asked if they used Daysy to get pregnant. Out of the whole cohort of 798 women, 69 (9.01%) answered the question with –yes-. Because according to the manufacturer, at least 3 cycles are required to precisely determine the fertile window, only woman with ≥ 3 (51 out of 69) cycles were considered for further evaluation. Among the 46 woman having sexual intercourse specifically on fertile (red) days, 18 (39.%) pregnancies were reported. In all cases, it took less than 1 year until the user conceived. For those who have not been pregnant while using Daysy, 6 (21.42%) already tried to conceive unsuccessfully before using the fertility monitor. A surprising observation was that of the women that wanted to become pregnant with the fertility monitor, 21.42% already had experience with NFP. Focusing all users, 40 out of 798 (5.01%) already had experience with NFP.

Regardless of for what the device was used, 90 out of 798 (11.2%) women reported to have experienced an abortion or miscarriage.

789 out of 798 (99%) of the questioned users would recommend the device to their friends. Surprisingly even the women who became unintentionally pregnant while using the fertility monitor would recommend the device to a friend.

One of the main hypotheses of this study was that through the additional use of an App, and thereby improved usability of the medical device, it is possible to enhance the usage safety as well as the method safety rate. For this we asked the participating women about the frequency as well as the apprehension effect of the app. We found, that 516 out of 798 (64.66%) use the additional app DaysyView in a daily, 239 (29.24%) weekly and 44 (5.51%) monthly manner (Fig. 5). On closer inspection, it is noticeable that the usage decreases with the number of cycles. Thus, after 4 cycles, 74% of the women spent time daily by using the app, this frequency drops by 20% to 51% after 13 or more cycles. In return, the weekly and monthly usage increases (Fig. 5).

In addition, 84% of the participants indicated that they achieved a better understanding of themselves and their cycle through the additional use of the app DaysyView.

Because sharing data is a key opinion of apps, we also ask if DaysyView users use this option. Interestingly, 506 out of 798 (63.2%) woman stated that they shared personal cycle data with their partner, friends or healthcare professionals.

1 year after the study was started (November 1st, 2016) the status of 776 (98%) DaysyView accounts is “Ready”, therefore DaysyView and the fertility monitor Daysy are still in use. 20 (2%) accounts and the corresponding serial numbers of the fertility monitor have been deleted from the server. Of the 778 remaining accounts, 618 (79%) were synchronized with the fertility monitor after June 1st 2017.

The contraceptive effectiveness of fertility monitors has already been demonstrated in different trials [3]. Since the Daysy algorithm is identical to the fertility algorithm of Babycomp and Ladycomp from Valley Electronics GmbH it was claimed that Daysy has a similar method as well as typical-use related PI to these products (The perfect-use related PI was not part of the previous study). The result of the method-related PI calculation obtained in the present study (0.63) is comparable to that advertised by the manufacturer and differs only a little from what is reported by Freundl and colleges (0.7) [5]. However, if the focus is on the typical-use related PI, it has improved from 3.8 to 1.25. There are three reasons for the enhanced effectiveness of the fertility monitor Daysy. (1) It has been over 20 years since the research of Freundl et al. was published. The software, as well as the shape of the device, has improved and the amount of historical data has increased. (2) The sociodemographic characteristic of the users has changed. In 1997 Freundl reported, that 69% of the participants are between 19 and 29 years old. In the current study, this fraction decreases by 11% while the fraction of 30–39 years-old increases by 10% to 37%. As it is already shown across all methods, contraceptive failures were significantly higher among younger woman (those younger than 25) than among older peers [22] (3). The calculated value for the cycle computers Babycomp and Ladycomp means they are highly effective in recognizing the different phases of the menstrual cycle and their usage is connected with the risk similar to that of other known contraceptive methods.

The perfect-use related PI was 0.8. Considering the life-table analysis it was 2.19%. This results in the possibility to compare the effectiveness of the fertility monitor Daysy with current other study results focusing on fertility awareness-based methods. Compared to other methods based on FABM, the Daysy fertility monitor positions itself at the upper end [23, 24]. However, for the remaining methods, the typical-use related pregnancy rate is worse than the perfect-use related pregnancy rate. Looking at the fertility monitor examined here, one notices that the typical- and perfect-use related pregnancy rates are much closer together. Basically, there are two possible explanations for improved result. (1) Most of the fertility awareness based methods require that female users observe or measure their signs of fertility and transmit them accordingly for evaluation. This can lead to interpretation or rounding errors which are reflected in the typical-use of the method. With a perfect-use related PI of 1 the algorithm used by an contraceptive mobile app has a very high safety [25]. If the typical-use related PI of 6.9 is considered, it becomes clear that the user in itself represents the greatest risk (which is our main hypothesis). The Fertility Monitor Daysy avoids this risk, women measure their basal temperature, which is stored on the device and automatically transmitted if necessary. This excludes rounding or interpretation errors. (2) The usability and understanding of a method plays a major role in a fertility monitoring device and its safe effective use. In the present study, users had the opportunity to use the additional app DaysyView as a display of their cycle data and thereby improve the usability of the medical device. Applications have the advantage of being available throughout the day, and they have the further ability to visualize complex cycle data in a simplified form. In the current study, the app DaysyView was used by 65% of the participants daily. Further, 84% of the participants indicated that they achieved a better understanding of themselves and their cycle through the additional use of the app DaysyView. The disadvantage of the app is that there is a kind of “wear-effect”. Thus, after 13 cycles, the daily use drops down to 51%. One reason could be that users are better acquainted with the method and they do not need a daily observation of their cycle any longer. Another reason could be that female users have a greater confidence in the fertility monitor, so that a visual check of the fertility status is no longer considered necessary.

One year after the survey started, 98% of the participants’ accounts were still active. If only the accounts that continue to be synchronized with the fertility monitor after 1st June 2017 are considered, it is still 79%. One way of explaining this discrepancy is, like already discussed above, that female users continue to use the fertility monitor, but no longer synchronize with the app. Another possibility is that, for pregnancy reasons, users do not use the fertility monitor during this time, and therefore do not synchronize further. Since participants only agreed during the period of the study (November 1st and December 31st 2016) that scientists had access to their data, this hypothesis could not be verified and will be part of further research.

The main reason for failure was unprotected (54%) as well as protected (46%) intercourse during the fertile phase (red days). As it is shown in Fig. 3 (inperfect-use) for the risk-taking women who had unprotected intercourse during the fertile time, the pregnancy probability increases significantly up to 10.82% after 13 cycles. This result is roughly comparable with the result that was found investigating the symptothermal method in 2007. It was reported that for this fraction who had unprotected intercourse during the fertile time, the pregnancy probability increases up to 7.5% per year [2].

In the current study, only 2 women indicated that they had unprotected intercourse during the infertile (green) phase and received an unintentional pregnancy. The probability of an unwanted pregnancy in this fraction is 2.2% which is slightly higher (n/s) than the incidence of women who have always used an additional contraceptive method (1.92%).

As already described: Through the digital analysis of temperature data, fertility monitors can reduce the risk of inaccurate or misinterpretation (as it is done by a computer) of fertility indicators and they can remind the user that a pregnancy risk exists on fertile (red) days but they cannot reduce the risk of the additional contraceptive methods or unprotected intercourse. It must be noted, that unwanted pregnancy is not a harm arising from the application of the device. It is a risk of FABM per se. Using fertility awareness-based methods means accepting this risk.

As discussed in the late 1970s, traditional BBT is not very effective in predicting the ovulation window to aid in conception [26]. This is because the method is only able to estimate the ovulation date after the fact. The software in Daysy uses statistical methods based on the previous cycles to provide a prognosis of the ovulation data upon which the users can act. It then compares the predicted date to the calculated ovulation date at the end of a cycle to update its model.

Daysy was used to achieve pregnancy in 9% of the participants. In 38% of cases there was a pregnancy within one year. Daysy was used on average for 8.5 months until the user conceived. Two studies showed that the fertility rates are perhaps higher than what we found in the current study. In the report published by Wang and colleagues, the authors found that − 50% became clinically pregnant in the first 2 cycles and > 90% in the first 6 cycles [27]. Another study analyzes the time to conception using FAB methods, and finds that for a fertile woman using FABM, the estimated cumulative probability of conception at 1, 3, 6, and 12 cycles was 38, 68, 81, and 92% [28].

The reason that our findings differ from the other results is that the largest share (70%) used Daysy for both, family planning as well as avoiding pregnancy (Fig. 1b). Thus, we don’t exactly know when they “start” family planning.

Obviously, many women are convinced by this technology: 99% would recommend the device to their friends. Even all of the women who became unintentionally pregnant while using the fertility monitor would still recommend the device.

We conclude that it is possible through the present technology of Daysy and the additional, optional use of DaysyView, to improve usability and enhance the usage safety as well as the method safety rate. Therefore, it seems that combining a specific biosensor-embedded device, which gives the method a very high repeatable accuracy, and a mobile application which leads to higher user engagement results in higher overall usability of the method.