Advice
Clinical and technical evidence
Clinical and technical evidence
A literature search was carried out for this briefing in accordance with NICE's interim process and methods statement. This briefing includes the most relevant or best available published evidence relating to the clinical effectiveness of the technology. Further information about how the evidence for this briefing was selected is available on request by contacting mibs@nice.org.uk.
Published evidence
Five studies are summarised in this briefing. These include data from 70,113 people (it is likely that there was substantial overlap between the people in the studies).
The clinical evidence and its strengths and limitations is summarised in the overall assessment of the evidence.
Overall assessment of the evidence
The 5 studies comprise observational data from people using the app who gave consent for their data to be shared for research purposes when downloading the app. The studies collected real-world data from people using the app. These people had no additional instructions or information from the investigators. There are no studies comparing Natural Cycles with other contraceptives and none that examine the effectiveness and accessibility of using the app to help plan a pregnancy. The life table analyses show Natural Cycles has a typical-use pregnancy rate of around 7%. This compares favourably with rates for calendar fertility-awareness methods (24%) or condom use (8%). NICE has endorsed an FPA resource comparing different contraceptive methods. Failure rate measures for different contraceptive methods are known to vary by country, age of person using the method, previous pregnancies, and type of previous contraceptive use. Grenfell et al. 2020 reports qualitative data of people in the UK about their experiences using a fertility-awareness app. This has not been summarised in the briefing because it does not report the efficacy of the app to prevent or help pregnancy.
Bull et al. (2019)
Key outcomes
On average, users entered data for 8 months. A 1-year typical use pearl index (PI) of 6.1 plus or minus 0.2 and a 13-cycle failure rate of 6.3% plus or minus 0.6%. People who had not recently used hormonal methods of contraception (n=9,381) had a PI of 5.1 plus or minus 0.3 and a 13-cycle failure rate of 5.2% plus or minus 0.7%. People who had recently used hormonal methods of contraception (n=6,950) had a PI of 7.5 plus or minus 0.4 and a 13-cycle failure rate of 8.1% plus or minus 1.0%. People that used condoms as a primary form of contraception before using Natural Cycles (n=2,411) had a PI of 3.5 plus or minus 0.5 and a 13-cycle failure rate of 3.6% plus or minus 1.0%. People that used the contraceptive pill before using Natural Cycles (n=4,023) had a PI of 8.1 plus or minus 0.6 and a 13-cycle failure rate of 8.7% plus or minus 1.3%.
Strengths and limitations
The study used data from all people who were paying to use the Natural Cycles app between 1 September 2016 and 30 October 2017 (and who had agreed to share data for research purposes). The study results are based on people engaging with the app; of the 16,331 people who met the inclusion criteria, 5,683 contributed at least 9 months. The cohort design was based on self-reported contraceptive use. The investigators made some assumptions about the pregnancy status if people stopped using the app. For example, people were assumed to be pregnant if they stopped during the late luteal phase or if they reported a high basal temperature. The study was funded by the company.
Kleinschmidt et al. (2019)
Intervention and comparator
Intervention: Natural Cycles.
Comparators: rhythm method; standard days method.
Key outcomes
The analysis compares the fraction of accurately and inaccurately predicted non-fertile days (described as green day [GD] and wrong green days [WGD], respectively) between Natural Cycles, rhythm method and standard day methods of fertility awareness. Natural Cycles' algorithms allocated 59% (95% confidence interval [CI] 58 to 59) GDs in cycle 12 and an average of 0.08% (95% CI 0.07 to 0.09) WDG. The rhythm method resulted in 46% (95% CI 45 to 46) GDs and 0.18% (95% CI 0.16 to 0.20) WGD over 12 cycles (excluding cycles 1 to 6), compared with Natural Cycles over 12 cycles (excluding cycles 1 to 6) 58% (95% CI 58 to 59) GDs and 0.07% (95% CI 0.07 to 0.08) WGDs. The standard day method over 12 cycles was 58% (95% CI 58 to 58) GD and 0.93% (95% CI 0.89 to 0.97) WGDs compared with Natural Cycles 56% (95% CI 56 to 56) GDs and 0.07 (95% CI 0.06 to 0.08) WGDs.
Strengths and limitations
The study used data from people paying to use the app collected prospectively between 1 September 2017 and 1 March 2019. The Natural Cycles app used was version 3.0. Basal body temperature measurements, menstruation dates and urinary luteinising hormone test results were recorded by the users directly into the app. People that had variable cycles, or cycles with fewer than 20 days did not take part in the study. This shows that the findings may not be generalisable to the wider population. The assumed 'true' ovulation day is based on retrospective placement according to the Natural Cycles algorithm and is associated with a small margin of error. The data were limited because of the lack of consecutive cycles with complete data entry, which reduced the number of cycles analysed in the study. The study was funded by the company.
Key outcomes
On average, users entered data into the app for 9.8 cycles. App data were used to calculate perfect and typical PIs. Typical-use PI: 6.8 (95% CI 6.4 to 7.2) and 13-cycle typical-use failure rate of 8.3% (95% CI 7.8 to 8.9). Perfect use PI: 1.0 (95% CI 0.5 to 1.5) and a typical-use failure rate of 8.3% (95% CI 7.8 to 8.9) over 13 cycles. Discontinuation over 12 months was 54%. Perfect use was defined as a cycle with no unprotected sex during a 'red day'.
Strengths and limitations
The study used data from all people paying to use the app between August 2014 and August 2017 (and who had agreed to share data for research purposes). Calculation of perfect use required that people record in the app when they had sex. It was optional for users to enter these data. The investigators relied on users to report pregnancies and give information by email. The investigators made some assumptions about pregnancy status if the person stopped using the app. For example, people were assumed to be pregnant if they stopped during the late luteal phase or if they reported a high basal temperature. The study was company funded.
Key outcomes
On average, people entered data into the app for 6.3 cycles. A total of 143 unplanned pregnancies happened during the study, giving a PI of 7.0 for typical use and a life table analysis gave a pregnancy rate of 7.5% per year (95% CI 5.9 to 9.1%). Ten of these pregnancies may have been because of the app incorrectly reporting a 'safe day' during the fertile window, giving a pregnancy rate of 0.5%.
Strengths and limitations
The study contains additional qualitative survey data, which were optional for users to input. It is unlikely that the data were collected across the whole group. Assumptions about pregnancy status were made in the absence of confirmation. The study was funded by the company.
Scherwitzl et al. (2015)
Study size, design and location
Retrospective study using data from 317 app users in Sweden.
Key outcomes
App users entered basal body temperature, ovulation test results and menstruation data. The mean delay between the first positive ovulation test to the temperature-based estimation of the ovulation day was 1.9 days. The length of the luteal phase varied on average by 1.25 days per person. Only 0.05% of non-fertile days were falsely attributed and found to be within the fertile window.