Effects of Technology-Based Contraceptive Decision Aids: A Systematic Review and Meta-Analysis

Abstract

Objective

To conduct a systematic review and meta-analysis of the effects of technology-based decision aids on contraceptive use, continuation, and patient-reported and decision-making outcomes.

Data sources

A systematic search was conducted in OVID MEDLINE, Cochrane Database of Systematic Reviews, CENTRAL, CINAHL, Embase, PsycINFO, and SocINDEX databases from January 2005 to April 2022. Eligible references from a concurrent systematic review evaluating contraceptive care were also included for review.

Study eligibility criteria

Studies were included if a contraceptive decision aid was technology-based (i.e., mobile/tablet application, web or computer-based) and assessed contraceptive use/continuation or patient-reported outcomes (knowledge, self-efficacy, feasibility/acceptability/usability, decisional conflict). The protocol was registered under the International Prospective Register of Systematic Reviews (CRD42021240755).

Study appraisal and synthesis methods

Three reviewers independently performed data abstraction and quality appraisal. Dichotomous outcomes (use and continuation) were evaluated with an odds ratio, while continuous outcomes (knowledge and self-efficacy) were evaluated with the mean difference. Sub-group analyses were performed for mode of delivery (mobile & tablet applications vs. web & computer-based) and follow-up time (immediate vs. > 1 month).

Results

This review included 18 studies evaluating 21 decision aids. Overall, there were higher odds of contraceptive use and/or continuation among decision aid users compared with controls (OR 1.27; 95% CI [1.05–1.55]). Use of computer and web-based decision aids was associated with higher odds of contraceptive use and/or continuation (OR 1.36; 95% CI [1.08–1.72]) compared to mobile and tablet decision aids (OR 1.27; 95% CI [0.83–1.94]). Decision aid users also had statistically significant higher self-efficacy scores (mean difference, 0.09; 95% CI [0.05–0.13]) and knowledge scores (mean difference, 0.04; 95% CI [0.01–0.07]), with immediate measurement of knowledge having higher retention than measurement after one month. Other outcomes were evaluated descriptively (e.g., feasibility, applicability, decisional conflict) but had little evidence to support a definite conclusion. Overall, the review provided moderate-level evidence for contraceptive use and continuation, knowledge, and self-efficacy.

Conclusions

The use of technology-based contraceptive decision aids to support contraceptive decision-making has positive effects on contraceptive use and continuation, knowledge, and self-efficacy. There was insufficient evidence to support a conclusion about effects on other decision-making outcomes.

Condensation:

Technology-based contraceptive patient decision aids can improve contraceptive use/continuation rates and increase knowledge and self-efficacy.

Introduction

Contraceptive choice is key to equitable contraceptive services and it supports reproductive autonomy and rights.^{1, 2} Preferences and values, shaped by health, social and personal factors, all play a role in patient-centered contraceptive decision-making.³ Characteristics of contraceptive methods, such as effectiveness, whether use is coitally dependent, ability to remove or stop the method independently (without a clinic visit), route of administration, and type or presence of hormones, may all be important to potential users.^{3, 4} Contraceptive choice is made more complex by the number of available options (currently 16 FDA-approved methods).⁵ This preference-sensitive decision requires that providers apply shared decision-making principles and patient-centered contraceptive counseling, prioritizing patient autonomy and ensuring reproductive health care equity.⁶

Patient decision aids are one way to facilitate shared decision-making that is grounded in patient preferences and values. Decision aids are most useful when each option has benefits and potential harms, and there is no clear best choice based on medical evidence.⁷ Decision aids can be a printed product, video, website, or mobile-based (i.e., mobile applications). They are designed to provide medical evidence about options in plain language and help patients form accurate perceptions of benefits and risks or potential harms.⁷ Overall, patients who use screening and treatment decision aids have increased knowledge about options and select methods better aligned with their preferences when compared to patients in usual care.^{7, 8} There is strong evidence that decision aids improve the decision-making process by reducing decisional conflict and better engaging patients to be part of the decision-making process.^{7, 8} With the expansion of technology in healthcare, decision aids have also moved towards using the latest technology (i.e., mobile, web-based, computer program), allowing information and decision-making to be more accessible to patients.⁹ Previous reviews of decision aids in contraceptive services have either focused on all modalities of decision aids (e.g., paper and technology-based) or decision aids within specific patient populations.^{10, 11} There is less evidence about technology-based contraceptive decision aids, a rapidly evolving area that creates accessibility to health information and decision-making.^{12, 13} The purpose of this systematic review was to examine how technology-based decision aids affect contraceptive use, self-efficacy, knowledge, and decision-making.

Objectives

We performed a systematic review to examine the following questions:

• What is the effectiveness of technology-based (e.g., website, mobile-based, etc.) contraceptive decision aids in improving the continuation or uptake of contraceptives? (KQ1)

• What are the impacts of technology-based contraceptive decision aids on patient-reported experiences (e.g., acceptability, patient knowledge, feasibility, decision process, etc.)? (KQ2)

Methods

We registered this systematic review in PROSPERO¹⁴ and performed according to the guidelines set by the Preferred Reporting Issues for Systematic Review and Meta-Analysis (PRISMA) for systematic review studies.¹⁵

Eligibility criteria

We included studies that focused on individuals with the potential for becoming pregnant without the use of contraception. This review focused on studies reporting on technology-based contraceptive decision aids (i.e., mobile, web-based, computer program). These decision aids incorporated at least one of the following: education about contraceptive methods, values clarification, or risk communication. Some decision aids also included screening for an individual’s need for contraceptive services and assessing medical risk factors relevant to specific contraceptive methods, pregnancy, and birth. Contraceptive methods include FDA-approved devices and medications in the United States as of 2020. Studies must compare the intervention to a control group or compare the intervention and measure one of two outcomes: 1) Continuation or uptake of contraceptive methods or 2) patient-reported experiences (i.e., feasibility, acceptability, improved knowledge, decision process). Studies can be conducted in various settings (e.g., health care clinics, health care systems, affiliated sites, educational facilities, pharmacy, telemedicine, homes) by a broad range of providers (e.g., physicians, nurses, pharmacists, counselors) or be self-initiated.

We developed abstract inclusion and exclusion criteria in accordance with the Methods Guide for Effectiveness and Comparative Effectiveness Reviews.¹⁶ The inclusion criteria were: randomized and non-randomized controlled clinical trials, observational studies with comparison groups including before/after studies, and systematic reviews of these study designs. Observational studies (i.e., before-after pilot studies) were included due to the iterative development cycle within patient decision aid research, which can provide great amounts of data involving patient-reported outcomes. Studies were excluded if they (1) were conducted with a different population focus (i.e., patients with uncommon conditions that would not be applicable to general populations); (2) were conducted with a decision aid that did not focus on FDA-approved devices and medications, was a paper-based or telephone decision aid, or decision aids that focused on use of contraceptive methods for non-contraceptive benefits (e.g., bleeding patterns); (3) did not include a comparator or control group; (4) reported a different outcome than included in our key questions (i.e., continuation/uptake of contraceptive methods or patient reported experiences); (5) conducted in settings that do not provide or refer patients to contraceptive counseling/family planning, or conducted in countries not rated “very high” on the 2018 Human Development Index (HDI);¹⁷ (6) study design other than controlled clinical trials with ≥ 30 participants, observational studies with comparison groups including before/after studies with ≥ 30 participants, qualitative studies relevant to KQ1 and KQ2, or systematic reviews of these excluded study designs; (7) article or systematic review covered by a more recent systematic review; (8) published in a language other than English.

Information Sources and Search Strategies

The initial search was guided by the Patient, Intervention, Comparison and Outcomes (PICO) model that we defined following standard protocol¹⁸ and was conducted by a trained medical librarian. Additional studies were identified in a concurrent systematic review during the abstract and full-text review phases to be included for this review.¹⁹ The electronic databases included OVID MEDLINE, Cochrane Database of Systematic Reviews, CENTRAL (Cochrane Central Register of Controlled Trials), CINAHL (Cumulative Index of Nursing and Allied Health Literature), Embase, PsycINFO, and SocINDEX databases. The search dates were restricted to January 2005 - April 2022 and only include studies relevant to current technology and contraceptive methods. In addition, reference lists of included articles were manually reviewed to identify other relevant studies. Similarly, a search was conducted (RG) to identify any articles related to conference abstracts and their respective reference lists. Sources for grey literature included reports produced by government agencies, health care provider organizations, or others. All search results were compiled into EndNote (Clarivate Analytics, Philadelphia, PA) and imported into Covidence (Veritas Health Innovation, Melbourne, Australia).

Study Selection

All authors reviewed titles and abstracts of articles, disagreements were resolved by consensus. Next, we independently reviewed the full text of articles included at the first stage. Disagreements were resolved by consensus.

Data Abstraction

We used a standardized data abstraction form to extract data into evidence tables that included study design, year, setting, country, sample size, eligibility criteria, population and clinical characteristics, details and characteristics about the decision aid, and results relevant to each key question. The articles were divided among authors (RG, KM, AB), abstracted independently, and verified by a second author. Discrepancies were discussed by the team and resolved by consensus. The evidence tables are available in Appendix Tables 1 & 2.

Risk of Bias Assessment

The risk of bias of individual controlled trials, systematic reviews, and observational studies was determined using predefined criteria developed by the US Preventative Services Task Force (USPSTF).²⁰ For randomized trials, we assessed randomization and allocation concealment methods, attrition, use of intention-to-treat methods, and blinding. For observational studies, we assessed factors such as patient selection methods, attrition, the accuracy of methods for measuring exposures, outcomes, confounders, and appropriateness of methods to address potential confounding. Each study was dual-reviewed for quality and applicability, with disagreements resolved by consensus. These criteria and methods were consistent with the USPSTF Procedure Manual.²¹ Studies were rated as “good,” “fair,” or “poor,” as specified by the quality assessment criteria. The overall strength of evidence was also assessed (RG) using the USPSTF Procedure Manual.²²

Data Synthesis

We pooled data for contraceptive use and continuation, patient knowledge, and self-efficacy outcomes and performed meta-analyses. For subgroup analyses, contraceptive use and continuation were stratified by technology modality, and studies measuring patient knowledge were stratified by follow-up time. Outcomes that had fewer than three trials with suitable data were assessed descriptively. We used an Empirical Bayes random-effects model for all meta-analyses using the “meta” package in STATA, version 16 (College Station, TX). Odds ratios were used to compare contraceptive use and continuation (dichotomous), while mean difference was used to compare our continuous outcomes (i.e., patient knowledge and self-efficacy). Statistical heterogeneity between studies was determined with the I² statistic and potential sources for heterogeneity were evaluated with subgroup and sensitivity analyses (i.e., Egger’s test for publication bias).²³ A p-value of <.05 indicated statistical significance for all analyses.

Results

Study selection

Of 1147 eligible records, 18 unique studies met the inclusion criteria (Figure 1).^24–41 We reviewed 86 studies at the full-text level and 22 for data abstraction. During data abstraction, we found incomplete data for four studies, thus their exclusion. Two included RCTs only provided data for the intervention group, so we only qualitatively evaluated the outcomes (e.g., patient knowledge and self-efficacy) from these studies.^{40, 42}

Figure 1.

Literature Search and Selection (PRISMA chart)

Study Characteristics

Of the 18 included studies, 17 were US-based^24–41 and one was conducted in the UK.³⁹ Fifteen were RCTs,^{24–30, 32–34, 36–40} one was a non-randomized controlled trial,³¹ one was a prospective cohort trial,³⁵ and one was a development study.⁴¹ Fifteen studies reported contraceptive uptake and continuation,^{25–28, 30–34, 36–40} ten assessed patient knowledge,^{26, 29–31, 33, 34, 37, 38} four explored self-efficacy,^{25, 33, 34, 40} four measured other decision-making outcomes, and two explored feasibility, acceptability, and usability.^{37, 41} Twelve studies only included women (18 years and older);^{24, 25, 27–29, 31, 32, 36–39, 41} three studies included women and adolescents/young adults (15 – 29),^{26, 30, 33} and three studies only included adolescents.^{34, 35, 40} Four studies only recruited women from racial minorities.^{24, 30, 34, 40} The most prevalent study setting was family planning/OB-GYN clinics.^{24, 27–31, 33, 35, 38, 40} Nine studies offered the decision aid through a mobile/tablet application,^{24, 26, 29, 30, 33–35, 38, 40} five used a web-based program,^{25, 27, 28, 31, 39} and three used a computer program.^{36, 37, 41} One study’s decision aid was built as a mobile app and a website.²⁶ Three of the web-based decision aids featured two different interventions within the same study.^{25, 27, 28}

Several studies grounded development of their decision aids in a variety of conceptual models or theories. The Theory of Planned Behavior was the most commonly used^{24, 29, 30, 33, 34} while eight studies did not report using a conceptual model or theory during decision aid development (Appendix Table 3).^{26–28, 32, 38–41}

Risk of bias of included studies

Articles displayed low risk of bias for adequate randomization, allocation concealment, stating eligibility requirements, reporting attrition and withdrawal rates, having similar groups at baseline, and analyzing people within groups. It was unclear whether a majority of the studies masked research participants, outcome assessors, or patients. Eight studies reported poor follow-up (less than 44%).^{25, 27–29, 36, 37, 39, 40} From our included studies, three studies were rated good quality,^{26, 33, 34} eleven were fair quality,^{25, 27–32, 37–40} and two were poor quality (Figure 2).^{24, 36} Two studies were not eligible for quality assessment due to study design.^{35, 41} Seven studies had good applicability,^{26–28, 33–35, 39} ten had fair applicability,^{24, 27–32, 37, 40, 41} and two were rated poor due to similarity of groups at baseline and ambiguity on masking several participants within the study.^{25, 36} Quality and applicability assessment results are found in Figures 2 & 3 and Appendix Tables 1 & 2.

Figure 2.

Risk of Bias by Study

Figure 3.

Risk of Bias Summary

Synthesis of results

Contraceptive use and continuation

Thirteen randomized control trials and one non-randomized control trial (N = 8486) were included for KQ1 (Appendix Table 1).^{25–28, 30–34, 36–40} Our meta-analysis included 11 studies;^{25, 26, 28, 30, 32–34, 36, 37, 39, 40} overall, technology-based application users had higher and significant odds of contraceptive use and/or continuation compared to control participants (OR 1.27; 95% CI [1.05–1.55]; I²=70.46%; p=0.00) (Figure 4). Three studies not eligible for meta-analysis due to insufficient or missing data had mixed results for increasing contraceptive use or continuation.^{27, 31, 38}

Figure 4.

Meta-analysis and sub-group analysis of contraceptive use and continuation

Mobile/Tablet Application decision aids:

Participants who used mobile or tablet decision aids reported higher but non-significant odds of use and/or continuation of contraception compared with control participants (OR 1.27; 95%CI [0.83–1.94]; I²=85.62%; p=0.10). Six RCTs^{26, 30, 32–34, 40} (N=4017) indicated higher rates of contraceptive use or continuation, with a decision aid containing a contraceptive assessment and tailored contraceptive recommendations showing the highest odds of contraceptive use for decision aid users recorded six months post-intervention (OR 5.54; 95% CI [1.70–18.06]). One study among adolescents with a decision aid focused on sexual and reproductive health information,³³ showed no difference in hormonal or LARC use at six weeks (aOR 1.00; 95% CI [0.81–1.22]). The replication study³⁴ (which only included Black or Latinx adolescents) also showed no change in hormonal or LARC use at six months (aOR 1.01; 95% CI [0.81–1.27]).

Computer & Web-based decision aids:

Use of computer and web-based decision aids appears to significantly increase odds of contraceptive use and continuation (OR 1.36; 95% CI [1.08–1.72]; I²=59.36%; p=0.02). Five RCTs^{25, 28, 35–37} (N=3999) demonstrated higher rates of contraceptive use or continuation, with a 50-question algorithm decision aid²⁸ having the greatest effect on effective method use three months post-intervention (aOR 1.86; 95% CI [1.44–2.41]). The four-month follow-up study²⁷ showed a high continuation rate for both intervention arms, with the tailored decision aid having a higher continuation than the control group (aOR 5.48; 95% CI [1.72–17.42]).

Knowledge

Eight studies (N=3700)^{26, 29, 30, 32–34, 37, 38} demonstrated statistically significantly higher contraceptive knowledge scores among decision aid users (MD: 0.09; 95% CI [0.05–0.13]; I²=53.86%; p=0.05) compared with control participants. Meaning, on average, decision aid users were nine percentage points more likely to retain correct contraceptive knowledge than their control counterparts. A LARC-focused mobile decision aid²⁹ had the largest difference (33 percentage points) in correct knowledge items (95% CI [0.08–0.57]) (Figure 5, Appendix Table 2). Two studies were not included in the meta-analysis due to incomplete data.^{35, 40}

Figure 5.

Meta and sub-group analyses of patient knowledge

The mean difference in correct knowledge items was larger when measured immediately after using a decision aid (MD 0.13; 95% CI [0.03–0.22]; I²=75.43%; p=0.03) and notably reduced when measured after a month (MD 0.06; 95% CI [0.02–0.10]; I²=0.00%; p=0.78) but remained significant at both time points (Figure 5).

Self-efficacy

Four studies (N = 3657)^{25, 33, 34, 40} examined self-efficacy (Appendix Table 2). The mean difference in self-efficacy scores was statistically significant between decision aid users and control participants (MD 0.04; 95% CI [0.01–0.07]; I²=31.91%; p=0.02) (Figure 6). Meaning, on average, decision aids users were four percentage points more likely to be confident in their contraceptive choice than those who do not use a decision aid. A mobile app decision aid that provided tailored contraceptive recommendations for Black and Latina adolescents demonstrated the largest difference in self-efficacy (MD: 0.09; 95% CI [0.03–0.15]).⁴⁰

Figure 6.

Meta-analysis of self-efficacy

Acceptability/Feasibility/Usability

Two studies (N = 323) evaluated the acceptability/feasibility/usability of computer-based decision aids (Appendix Table 2).^{37, 41} Both studies examined decision aids developed to increase patient participation in contraceptive counseling sessions and improve provider knowledge of a patient’s specific needs. The studies’ results suggest the decision aids were acceptable and implemented into the clinical setting with few issues. In both studies, the decision aids were viewed positively by clinicians and patients.

Other Decision-making Outcomes

Two RCTs evaluated decision-making outcomes.^{24, 26} One study found a slight improvement in decision balance and another study found no difference in decisional conflict.

Sensitivity Analyses

We completed sensitivity analyses for our meta-analyses to explore whether publication bias contributed to high heterogeneity. These analyses did not show publication bias for contraceptive use and continuation (p=0.1274), patient knowledge (p=0.3001), or self-efficacy (p=0.2726) (Appendix Figures 1–3).

Comment

Principal Findings

Moderate-quality evidence found that using technology-based contraceptive decision aids is associated with higher rates of contraceptive use and/or continuation, contraceptive knowledge, and self-efficacy (Table 1). The meta-analysis results suggest that computer and web-based decision aids are associated with 36% higher odds of use and continuation compared to 27% higher odds (non-significant) for mobile and tablet applications. On average, decision aid users had higher contraceptive knowledge immediately after the decision aid intervention than after one month. Decision aid users were four percentage points more likely to be confident in their contraceptive choice (self-efficacy) than those who do not use a decision aid. Other outcomes found in this review were evaluated descriptively (e.g., feasibility, applicability, decisional conflict) but had little evidence to support a definite conclusion.

Table 1.

Summary of Evidence

Outcome	Study Design & Participants(N)	Summary of Findings	Consistency; Precision	Strength of evidence; Applicability
Use and Continuation (KQ1)	11 RCTs (8140)	Higher rates of contraceptive use and continuation in most studies for decision aid users (OR 1.34; 95% CI [1.06–1.68]). When stratified by mode of delivery, mobile & tablet applications showed a non-significant effect (OR 1.39; 95% CI [0.80–2.40]), and computer and web-based decision aids showed a statistically significant effect (OR 1.44; 95% CI [1.13–1.85]).	Inconsistent; Precise	Moderate; Moderate
Knowledge (KQ2)	7 RCTs & 1 Non-RCT, (3700)	All studies show higher scores for decision aid users vs. control counterparts (MD 0.09; 95% CI 0.05–0.13]). When stratified by follow-up time, decision aid users’ knowledge scores were higher when measured immediately (MD 0.13; 95% CI [0.03–0.22]). After one month, knowledge scores lower in effect but remain significant (MD 0.06; 95% CI [0.02–0.10]).	Consistent; Precise	Moderate; Moderate
Self-Efficacy (KQ2)	4 RCTs (3657)	All studies show statistically significant higher self-efficacy scores vs. the control (MD 0.04; 95% CI [0.01–0.07]).	Consistent; Precise	Moderate; Moderate

Comparison with Existing Literature

Few studies have evaluated the effectiveness of technology-based decision aids on contraception use, knowledge, self-efficacy, and decision-making. Our systematic review focused on studies of technology-based contraceptive decision aids dating back to 2005. A 2020 systematic review examined the use of any modality patient decision aid in obstetrics and gynecology more broadly and included eight studies focusing on contraceptive decision-making, a subset of which reported a decrease in decisional conflict and increase in knowledge.⁴⁴ We found similar results for knowledge but could not establish a trend for decision quality outcomes like decisional conflict. A 2021 systematic review that focused on adolescent and young adult (AYA) tailored decision aids, using all modalities that contained multiple contraceptive methods reported the aids improved knowledge but could not identify any trend for change in contraceptive use.¹¹ Our study found mixed results in adolescent populations related to knowledge scores but higher rates of contraceptive use and higher self-efficacy scores for AYA decision aid users. Our review also found statistically significant higher odds of use and/or continuation and higher scores in contraceptive self-efficacy for decision aid users of all ages. None of the aforementioned studies focused on contraceptive decision making with racial minorities, but a small subset of studies included in our review found higher rates of contraceptive use and self-efficacy.^{24, 30, 34, 40}

Strengths and Limitations

There were a few limitations of this study, some due to the nature of the literature and its impact on our ability to quantitatively synthesis the evidence. First, there were no standards for intervention development or outcome instruments. This required standardization for a portion of the data used in the meta-analyses. Use of validated instruments to measure contraceptive use/continuation and contraceptive knowledge could improve the results seen with the meta-analyses (a validated instrument already exists for self-efficacy⁴⁵). The use of standardized guidelines for decision aid development (like the International Patient Decision Aid Standards (IPDAS)) could enhance the quality and effectiveness of patient decision aids, ensuring evidence-based and health literate information be available to patients.⁴⁶

Second, for studies with multiple interventions,^{25, 28} each decision aid was treated as an individual study, causing inter-study correlation within our meta-analyses. This analysis decision could have contributed to high heterogeneity for certain sub-groups (I²), but other factors (i.e., different participants, clinical settings) can also contribute to the high statistic. To control for the correlation, we utilized a Bayesian correction.

Third, most of the decision aids were not tailored for vulnerable populations, limiting the applicability of the outcomes to the general US population. The studies have a diverse set of clinical settings and a subset of studies solely recruited racial minorities. Four studies^{24, 30, 34, 40} (N=2117) focusing on Black and/or Latina populations showed higher rates of contraceptive use and higher self-efficacy scores, but more high-quality RCTs are needed to confirm the preliminary trend.

Conclusions and Implications

Future studies should focus on following advancements in digital health modalities to support contraceptive decision-making, specifically evaluating decision process outcomes. Improvement in decision-process outcomes (i.e., decisional conflict) can increase the likelihood that patients choose a contraceptive choice they are satisfied with and continue to use.^{47, 48} Usability and acceptability may come easier with introducing decision aids with modalities patients already use.⁴⁹

Results of this systematic review show that the use of technology-based decision aids increases contraceptive use and/or continuation compared to control groups in clinical and online settings. Technology-based contraceptive decision aids also increase knowledge and self-efficacy. Limited evidence suggests the use of decision aids is feasible and acceptable to providers and patients. Other decision-making outcomes (i.e., decisional conflict and decision balance) had little evidence to support a definite conclusion. These results suggest that technology-based decision aids to support contraceptive decision-making are effective tools in clinical practice. Technology-based contraceptive decision aids are a promising and feasible intervention to enhance patient-centered contraceptive care.

AJOG at a Glance:

1. To explore the effects of technology-based decision aids on contraceptive use, continuation, patient-reported and decision-making outcomes.

2. Use of technology-based contraceptive decision aids demonstrated higher contraceptive use and continuation rates, and higher knowledge and self-efficacy scores.

3. This systematic review and meta-analysis show how the use of technology-based contraceptive decision aids can improve outcomes in reproductive health.

Appendix

Appendix Table 1.

Studies of Technology-based Contraceptive Decision Aids on Contraceptive Use and Continuation

Author, Year	Mode of Delivery	Setting (N sites)	Population (Analysis)	Intervention	Comparison	Outcome Measure	Results	Quality; Applicability
Chuang, 2019	Website	NR	Women (733)	RLP: CDC program interactive tool with individualized contraceptive information and family planning decision support; RLP+ group: RLP + contraceptive action planning tool	Viewed website for standard information about contraceptives	Contraceptive use and continuation	Any use: RLP vs. control: OR 0.81 (95% CI: 0.52–1.28) Rx use: RLP vs. control: OR 0.86 (95% CI: 0.70–1.07) High adherence: RLP vs. control: OR 1.01 (95% CI: 0.79–1.29)	Fair; Poor
Delhendorf, 2019	Website and Mobile App	Medical Clinics (4)	Adolescents, Women (749 (baseline), 645 (7 months))	Interactive contraceptive decision support tool with educational modules, personalized recommendations based on patient preferences for use in visit with provider	Standard contraceptive counseling with provider	Choice of contraceptive method; Contraceptive continuation	Highly effective contraception choice at baseline: 38.1% vs 35.2%; OR 1.06 (95% CI: 0.69–1.65), p = .78 Continuation at 7 months: 56.6% vs. 59.6%; OR 0.89 (95% CI: 0.65–1.22), p = .46	Good; Good
Garbers, 2012a	Website	Public Family Planning Clinics (2)	Women (1983)	Tailored: 50-question online algorithm, personalized contraceptive report, and contraceptive counseling; Generic: 50-question online algorithm, generic education about contraceptives and contraceptive counseling	Received clinic contraceptive handout with usual care	Effective contraceptive use	Use at 6 weeks: Tailored Intervention: 1.63 (95% CI: 1.28–2.07), p<0.001, Generic Intervention: 1.86 (95% CI: 1.44–2.41), p<0.001	Fair; Fair
Garbers, 2012b	Website	Family Planning Clinic (2)	Women (224)	Tailored: 50-question online algorithm, personalized contraceptive report, and contraceptive counseling; Generic: 50-question online algorithm, generic education about contraceptives and contraceptive counseling	Received clinic contraceptive handout with usual care	Continuation of effective contraceptives	Continuation at 4 months: Tailored: 5.48 (95% CI: 1.72–17.42), p=0.004; Generic: 1.31 (95% CI: 0.58–2.98), p=.518	Fair; Good
Hebert, 2018	Mobile App	Family Planning Clinics (4)	Women, Adolescents (207)	Theory-based mobile app (miPlan) on contraceptive methods with information and patient experience videos	Standard clinic visit	Contraceptive use and continuation	Use at 3 months: IUD: 1/88 vs. 2/78, p = 0.60, Implant: 6/88 vs.1/78, p = 0.12, Any LARC: 7/88 vs. 3/78, p = 0.34	Fair; Fair
Koo, 2017	Web-based Tool	Family Planning Clinics (County health department, Planned Parenthood) (2)	Women (126)	Computer-based tool focused on patient centered contraceptive counseling	Contraceptive counseling as usual	Method chosen at clinic	Immediate use of any contraceptive: IUD/Implant: 11/100 vs. 19/100, Injectable/ring/patch: 26/100 vs. 26/100, Pill: 63/100 vs. 53/100	Fair; Fair
Madden, 2020	Tablet App	OB-GYN clinics (2)	Women (241)	Tablet-based contraceptive decision aid eliciting preferences about reversible contraception and provides information about patient-preferenced options.	Completed generic questionnaire and received a non-tailored reproductive health care handout	Contraceptive use post-visit	Chose new contraceptive method: 51.6% (83/161) vs. 48.6% (39/80), p=0.66	Fair; Fair
Manlove, 2020	Mobile App	Online	Women, Young adults (1124)	Interactive app with comprehensive sexual and reproductive health information	App focused on health eating, exercise, sleep, and general health	Current hormonal/LA RC use; Use at last sex	Use at 6 weeks: 48.9% vs. 49.1%, p=.945 Use at last sex in the past 6 weeks: 49.1% vs. 51.7%, p=.379	Good; Good
Manlove, 2021	Mobile App	Online	Adolescents (1124)	Interactive app with comprehensive sexual and reproductive health information	App focused on health eating, exercise, sleep, and general health	Current hormonal/LA RC use; Use at last sex	Use at 6 months: 37.2% vs. 36.7%; p=0.858 Use at last sex at 6 months: 43.4% vs. 40.8%; p=0.508	Good; Good
Peipert, 2008	Computer Program	Hospitals (2); Planned Parenthood Clinic (1)	Women (346)	Computer-assisted, multimedia program on dual condom and contraceptive use	Computer-based, information on contraceptive methods; sample condom	Current contraceptive use	Dual method use at 24 months: 86/272 (32%) vs. 71/270 (26%), aHRR = 1.70; (95% CI: 1.09–2.66) Consistent condom use at 24 months: 124/272 (46%) vs. 124/270 (46%), aHRR = 1.26; (95% CI: 0.88–1.52)	Poor; Poor
Schwarz, 2013	Computer Program	Urban Acute Care Settings (Urgent Care, ER) (4)	Women (198)	Computer module screening for contraindications to OC: education on contraception; and offered the option to request prescription of OC, ring, or patch	Computer module screening women for chlamydia infection	Receipt of OC prescription; Contraception use at last intercourse; Use of prescribed OC at last intercourse	Received Rx at 3 months: 16.2% (19/117) vs 1.3% (1/80); p<0.001 Used Rx at last intercourse at 3 months: 47.0% (55/117) vs 43.8% (35/80), p=0.95; aOR 1.41, (95% CI: 0.62–3.19), NS Use at last intercourse at 3 months: 70.9% (83/117) vs 65.0% (52/80), p=0.91; aOR 1.37, (95% CI: 0.33–5.68), NS	Fair; Fair
Sridhar, 2015	Mobile App	OB/GYN Center (1)	Women (120)	Application on tablet displays information on birth control methods	Received health information from a health educator	Use of effective methods	Contraceptive method of choice: p = 0.753 Very effective methods: 52% (31/60) vs. 57% (34/60), Effective methods: 28% (17/60) vs. 28% (17/60), Less effective methods: 20% (12/60) vs. 15% (9/60)	Fair; Fair
Stephenson, 2020	Website	Sexual and reproductive health service (2), Abortion service (1), Community pharmacy (1), Maternity service (1), General practice (1)	Women (739)	A self-guided website that presents a number of contraceptive options in response to patient preferences and concerns, provides information and videos about contraceptive methods and patient experiences.	Received contraceptive care and viewed the website at the end of the clinical trial	LARC method in use at 6 months	LARC method use at 6 months: 31.4% (93/364) vs. 31.0% (113/375), aOR 0.99 (95% CI: 0.66 – 1.49); p=0.98	Fair; Good
Tebb, 2021	Mobile App	School Based Health Center (18)	Adolescents (676)	Ipad Air survey with Health-E You/Salud iTu app with tailored contraception recommendations; printout to review with provider	Baseline survey on Ipad	Contraceptive use	Received non-barrier contraceptive at visit: 74.0% (236/319) vs. 50.7% (221/436), aOR 1.66 (95% CI: 0.73–3.78) Use in past 3 months, at 6 months: 62.7% (185/295) vs. 43.8% (166/379); aOR 5.54 (95% CI: 1.70–18.06)	Fair; Fair

Abbreviations: (a)OR – (adjusted) Odds Ratio; aHRR – adjusted Hazard Risk Ratio; ER – Emergency Room; IUD – Intrauterine device; LARC – Long acting reversible contraceptive; NA – Not applicable; NR – Not Reported; NS – Not significant (unless otherwise stated, cutoff is p>0.05); OB-GYN – Obstetrics & Gynecology; OC – Oral contraceptives; Rx - Prescription

Appendix Table 2.

Studies of Technology-based Contraceptive Decision Aids on Patient-Reported Outcomes

Author, Year	Mode of Delivery	Setting (N sites)	Population (Analysis)	Intervention	Comparison	Outcome Measure	Results	Quality; Applicability
Akinola, 2019	Mobile App	Community Based Family Planning Clinics (4)	Women (110)	Theory-based mobile app (miPlan) on contraceptive methods with information and patient experience videos and contraceptive counseling	Standard clinic visit	Contraceptive self-efficacy; Positive and Negative Decision Balance	Self-efficacy: 18.5 vs 17.6, p=.33 Positive decision balance: 22.8 vs 23.2, p=.042 Negative decision balance: 12.1 vs 12.9, p=.64	Poor; Fair
Chuang, 2019	Website	NR	Women (733)	RLP: CDC program interactive tool with individualized contraceptive information and family planning decision support; RLP+ group: RLP + contraceptive action planning tool	Viewed website for standard information about contraceptives	Contraceptive self-efficacy	Change in self-efficacy score: RLP vs. control: OR 0.36 (95% CI: −0.41–1.12), RLP+ vs. control: OR 0.42 (95% CI: −0.32–1.20),	Fair; Poor
Delhendorf, 2019	Website and Mobile App	Medical Clinics (4)	Adolescents, Women (749 (baseline), 645 (7 months))	Interactive contraceptive decision support tool with educational modules, personalized recommendations based on patient preferences for use in visit with provider	Standard contraceptive counseling with provider	Decisional conflict in contraceptive choice	Decisional Conflict Overall score: 25.4 vs. 22.6, OR 1.18 (95% CI: 0.80–1.74), p = .41 Correct knowledge on all IUD questions: 36.1 vs. 19.1, OR 2.45 (95% CI: 1.75–3.49), p < .01	Good; Good
Gilliam, 2014	Mobile App	Family Planning Clinics (3)	Women (52)	Mobile app used to learn about and answer method specific questions about LARC	Standard contraceptive counseling care	Increased contraceptive knowledge	Knowledge of contraceptive awareness: Intervention: 1 vs 1.5, p = .0001	Fair; Fair
Hebert, 2018	Mobile App	Family Planning Clinics (4)	Women, Adolescents (207)	Theory-based mobile app (miPlan) on contraceptive methods with information and patient experience videos	Standard clinic visit	Contraceptive knowledge	Knowledge at 3 months: Total Score: 1.5 vs. 1.8, p = 0.04	Fair; Fair
Koo, 2017	Web-based Tool	Family Planning Clinics (County health department, Planned Parenthood Health Center) (2)	Women (126)	Computer-based tool focused on patient centered contraceptive counseling	Contraceptive counseling as usual	Contraceptive knowledge; Patient-centeredness of counseling	Mean number of birth control methods known: 11.1 vs. 10.7, p<.001 Mean rating of patient-centeredness of counseling: 3.9 vs. 3.7, p<.05	Fair; Fair
Manlove, 2020	Mobile App	Online	Women, Young adults (1124)	Interactive app with comprehensive sexual and reproductive health information	App focused on health eating, exercise, sleep, and general health	Contraceptive knowledge; Birth control self-efficacy	Knowledge at 6 weeks: 51.5% vs. 44.5%, p < 0.05 Self-efficacy at 6 weeks: 67.3% vs. 61.5%, p = .025	Good; Good
Manlove, 2021	Mobile App	Online	Adolescents (1124)	Interactive app with comprehensive sexual and reproductive health information	App focused on health eating, exercise, sleep, and general health	Contraceptive knowledge; Birth control self-efficacy	Knowledge at 6 months: 50.6% vs. 47.1%; p=0.096 Self-efficacy at 6 weeks: 52.9% vs. 51.5%; p=0.645	Good; Good
Mesheriakova, 2017	Tablet App	School Based Health Centers (3)	Adolescents (120)	App addresses all contraceptive methods using videos; has an interactive contraceptive choice module factoring in the patient’s childbearing timeline.	NA	Sexual Health Knowledge	Sexual Health Knowledge (pre-post): 58% vs. 79%, p ≤ 0.002	NA; Good
Schwarz, 2013	Computer Program	Urban Acute Care Settings (Urgent Care, ER) (4)	Women (198)	Computer module offers contraception education, OC contraindications screening and prescription of OC, ring, or patch	Computer module screening women for chlamydia infection	Acceptability of computer-assisted provision	Acceptability at 3 months: 95% said the program was easy to use, 85% said they would recommend the program to a friend, 65% preferred discussing family planning needs with a healthcare provider	Fair; Fair
Sridhar, 2015	Mobile App	OB/GYN Center (1)	Women (120)	Application on tablet displays information on birth control methods	Received health information from a health educator	Knowledge about chosen method	Mean knowledge scores: 5.35 vs. 5.56, p = 0.30	Fair; Fair
Tebb, 2021	Mobile App	School Based Health Centers (18)	Adolescents (676)	IPad Air survey with Health-E You/Salud iTu app with tailored contraception recommendations; printout to review with provider	Baseline survey on IPad	Contraceptive knowledge; Contraceptive self-efficacy	Knowledge (Intervention only): 3.3 to 4.6; p <0.001 Self-efficacy at 6 months: 26.1 vs. 23.4, aOR 1.58 (95% CI: 0.38–2.77)	Fair; Fair
Wilson, 2014	Computer Tool	County Health Department Clinic (2), Planned Parenthood Health Center (1)	Women (125 patients, 7 providers)	Computer-based contraceptive counseling aid with a questionnaire and birth control guide for interactive learning	NA	Feasibility; Usability/acceptability	Feasibility: Average time = 14 minutes Usability/Acceptability: Did not understand all the questions: 3.2%; The questionnaire was too long: 9.6%; Unneeded questions: 16.8%; Important questions asked: 99.2%; Information guide was helpful: 100%; Learned a lot about methods: 96.8%;	NA; Fair

Abbreviations: (a)OR – (adjusted) Odds Ratio; ER – Emergency Room; NA – Not applicable; OB-GYN – Obstetrics & Gynecology; OC – Oral contraceptives

Appendix Table 3.

Conceptual Models and Theories used to develop contraceptive decision aids

Conceptual Model/Theory	Number of Studies*
Theory of Planned Behavior	5
Social Cognitive Theory	3
Transtheoretical Model of Behavior Change	3
Social Learning Theory	2
Chronic Care Model	1
Human Centered Design	1
Self-Efficacy Theory	1
Self-Regulation Theory	1
No Model/Theory Reported	8

^*Some studies used more than one model to develop their decision aid.

Appendix Figure 1.

Sensitivity analysis for contraceptive use and continuation

Appendix Figure 2.

Sensitivity analysis for patient knowledge

Appendix Figure 3.

Sensitivity analysis for self-efficacy