Design, protocol information, and patient involvement

We performed a systematic review, which was written according to the PRISMA 2020 statement. The study protocol was registered at PROSPERO (CRD42020187354), and there were no subsequent changes to the protocol. Raw data of the included studies can be accessed at https://figshare.com/s/a62bcb6af46d3e692327. Of note, patients and other relevant actors were not involved in this review

Eligibility criteria

We performed a systematic review that assessed all RCTs that were published as original papers in scientific journals, which compared the effect of different types of peri-abortion (either shortly before or shortly after abortion) contraceptive counseling interventions (which could involve material and/or human resources). We defined standard counseling as the intervention regularly practiced in the study context and enhanced counseling as the new intensified strategy considered for the trial. No restrictions in language or publication date were applied.

Information sources and search strategy

We performed a literature search of four sources: PubMed, Central Cochrane Library (CENTRAL), Scopus, and Google Scholar. Since Google Scholar sorts its results starting with those that have the best match with the search terms, we consider that evaluating the first 100 results would include all relevant studies on the subject in this repository, a methodology that has been used in previous systematic reviews [8–11].

We searched the articles using the terms “counseling” (A), “abortion” (B), and “randomized controlled trial” (C), with the following syntax: A AND B AND C. The detailed search strategies for each search source are available in S1 Table.

Selection process

The search was performed in two steps: 1) systematic search in four databases and 2) review of all references of the studies included in Step 1.

For step 1, we performed a literature search in June 2021, downloaded all results to an EndNote X8 document, and eliminated duplicated articles using this software. Subsequently, we assessed the titles and abstracts of each reference to identify potential studies for inclusion. Lastly, we assessed the full text of these potential studies to determine their eligibility.

For Step 2, we reviewed all references of studies that were included in Step 1 and collected new articles that met the inclusion criteria.

Both steps were performed independently by two authors. When disagreements were found, they were discussed by all authors and resolved by consensus.

Data collection process

Two authors independently extracted the following information for the included studies into a Microsoft Excel sheet: author, year of publication, year of collection, title, country, population (inclusion and exclusion criteria), setting, peri-abortion contraceptive counseling given to the intervention and comparator groups (using the Template for Intervention Description and Replication (TIDieR) data extraction tool) [12], and results for all outcomes assessed in the studies. In case of disagreements, the full-text articles were reviewed again by all authors.

Outcomes of interest

We evaluated the effect of contraceptive counseling interventions on contraceptive use (when an effective contraceptive method was used for more than three months), contraceptive uptake (when selected immediately after counseling), and obstetric event recurrence.

We considered effective contraceptive methods as any of the following: oral contraceptives, patch, ring, monthly injectable, quarterly injectable, condom, implant, vaginal ring, contraceptive patch, intrauterine device (IUD), intrauterine system (IUS), and sterilization or vasectomy. We considered long-acting reversible contraceptives (LARC) as any of the following: IUD, IUS/hormonal IUD, and implant. We considered obstetric events as either an unintended pregnancy or another induced abortion after counseling.

Evaluation time was considered as the moment in which the result was evaluated. This definition was applied for the outcomes of “use” and “uptake.” Moreover, the follow-up time was considered as the maximum time that the participants were followed in each study. This definition was applied for the outcome of “occurrence of an obstetric event.”

Risk-of-bias assessment

To evaluate the risk of bias of included RCTs, we used the Cochrane risk-of-bias tool for randomized trials [13], which assesses the risk of bias in seven domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcomes assessment, incomplete outcome data, selective reporting, and other potential threats to validity.

Synthesis methods

Whenever more than one study assessed a similar PICO (Population, Intervention, Control, and Outcome) question and the frequency of study events was greater than one in each group, we performed meta-analyses to summarize their results.

We assessed heterogeneity using an I² statistic, and heterogeneity was arbitrarily categorized using cutoff points agreed by the authors as not important (I² < 40%), moderately significant (I² = 40%–75%), and considerable (I² > 75%) [13]. We considered it appropriate to use random-effects models due to the overall heterogeneity of the interventions received by the participants in each study [14]. The data were processed using Review Manager 5.4 software.

Publication bias was not statistically assessed since the number of studies pooled for each meta-analysis was less than ten [14].

Certainty assessment

To assess the certainty of the evidence for each outcome, we used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology [15, 16], which evaluates the study design, risk of bias, inconsistency, indirectness, imprecision, and publication bias.

Study selection

We found 908 articles in the database search. After duplicates were removed, we screened 681 articles, from which 60 underwent full-text review. Then, 50 articles were excluded (detailed reasons for excluding these records are shown in S2 Table), and 10 studies were included [17–26]. Subsequently, we screened the 284 references of these 10 articles, from which one new study was included [27], for a total of 11 included studies (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flow diagram (study selection).

https://doi.org/10.1371/journal.pone.0260794.g001

Study characteristics

Studies’ characteristics are summarized in Table 1 and presented at length in S3 Table. Moreover, the characteristics of the interventions in each study were identified in detail with the help of the TIDieR tool (S5–S15 Tables).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of included studies.

https://doi.org/10.1371/journal.pone.0260794.t001

The 11 included studies were published between 2004 and 2017 but performed between 1999 and 2016. All were published in full-text in english. Three studies were performed in the USA [19, 20, 26], while the other eight were conducted in different countries. Regarding the continents, of 11 studies, four were performed in North America [19, 20, 23, 26], three in Europe [17, 22, 24], two in Asia [25, 27], one in East Africa [21], and one in South America [18].

Of the 11 included studies, seven showed the age means, ranging from 22.8 to 26.8 years. Furthermore, seven studies reported the frequency of previous abortion in their populations, ranging from 27.6% [18] to 50.5% [19]. The sample size ranged from 43 to 2336, and the follow-up period ranged from immediately to 16 weeks.

The included studies assessed two comparisons: nine studies compared enhanced contraceptive counseling and standard contraceptive counseling, provided to women before or after an abortion, while the other two studies compared abortion provision and contraceptive counseling, provided by physicians versus non-physician (nurses in one study and midwives in the other study) (Table 1). Furthermore, of 11 studies, six reported that the intervention was adjusted to the needs of each patient [17, 18, 20, 22, 25, 26], three [23, 24, 27] did not specify it, and two were not [19, 21] (more details in S5–S15 Tables).

Risk of bias in studies

Among the 11 included studies, only one had a low risk of bias in all Cochrane Tool items (Whitaker 2016) [26], while nine adequately generated the randomization sequence, seven appropriately concealed allocation, three blinded participants and personnel, two blinded outcome evaluators, seven did not report a significant number of missing outcome data, and four had a protocol available where no selective outcome reporting was found (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Risk of bias included studies (cochrane tool).

https://doi.org/10.1371/journal.pone.0260794.g002

Summary of the results

We found that nine studies compared standard care and enhanced contraceptive counseling [17–20, 22, 24–27], while the other two studies compared contraceptive counseling provided by a physician versus non-physician [21, 23].

Regarding the comparison of standard and enhanced contraceptive counseling, the definition of each outcome differed across studies, as detailed in S4 Table. Moreover, each study had a particular way to enhance their regular contraceptive counseling, such as the addition of pre-abortion counseling sessions [17], enhancement of contraception provision [24], personalized contraceptive counseling [22], use of audiovisual material [19, 20], several stage counseling [18], mobile phone interventions [25], and motivational interview [26], as detailed in S5–S15 Tables.

Pooled analysis showed that, compared to standard care, enhanced contraceptive counseling might increase the incidence of effective contraceptive method use (eight RCTs; relative risk [RR], 1.12; 95% CI, 1.09–1.16; I2 = 93%), may have little to no effect on the incidence of LARC use (three RCTs; RR, 1.25; 95% CI, 0.68–2.29; I2 = 68%), may have little to no effect on the incidence of effective contraceptive method uptake (five RCTs; RR, 1.06; 95% CI, 0.98–1.15; I2 = 84%), and may have little to no effect on the incidence of obstetric event occurrence (three RCTs; RR, 0.91; 95% CI, 0.57–1.47; I2 = 63%); however, the evidence is very uncertain for these results. Meta-analyses are shown in Fig 3, and the Summary of Findings table is shown in Table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Meta-analyses of studies that compared peri-abortion enhanced contraceptive counseling (intervention group) vs standard contraceptive counseling (control group).

https://doi.org/10.1371/journal.pone.0260794.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Summary of Findings (SoF) table.

https://doi.org/10.1371/journal.pone.0260794.t002

Regarding the comparison of counseling provided by physicians and non-physicians, two RCTs were included: Olavarrieta 2015 (N = 884) and Makenzius 2017 (N = 803) [21, 23]. Due to the differences in the control group of the studies, we did not perform a meta-analysis.

Olavarrieta 2015 [23] compared counseling provided by physicians with counseling provided by nurses and did not find significant differences in the proportion of women prescribed contraceptives (nurse group: 99.1%, physician group: 98.7%), in the type of contraceptive prescribed, and in the proportion of women leaving the facility with at least one contraceptive method (96.7% vs 97.3%). However, the use of an intrauterine device was higher in the “physician group” than in the “nurse group” (31.3% vs 24.0%), while condom use was higher in the “nurse group” than in the “physician group” (19.2% vs 10.6%).

Makenzius 2017 [21] compared counseling provided by physicians and counseling provided by midwifes, and did not find significant differences in the proportion of women who received contraceptive counseling (midwife group: 98%, physician group: 98%), and in the accepted contraceptive method (74% vs 77%).

Contraceptive use

Regarding the contraceptive use outcome, the meta-analysis showed that the enhanced counseling group had a higher incidence of this outcome compared with the standard counseling group. However, study results were heterogeneous. Four studies had a high uptake incidence in their control group (Bender, 85.2%; Langston, 73.5%; Schunmann, 98.6%; Zhu, 89.4%), so the enhanced counseling group could not have a much higher incidence; therefore, it was not surprising that the intervention did not seem to be beneficial.

As enhanced interventions seemed to have a higher effect on contraceptive use, six studies primarily viewed personalized counseling as enhanced antisense counseling [17, 18, 20, 22, 25, 26]. Personalized counseling consisted mainly of providing information about different contraceptive methods [17, 20, 24, 25] for approximately 30 min [18, 22]. However, interventions were moderately heterogeneous across studies and varied in the provision of contraceptive methods, from studies providing free contraception [18, 27] to others providing a 3-month provision only for specific methods like pills, IUD, and implants [24].

The study that had more weight (45.4%) in the forest plot was Zhu’s study, a study conducted in China, in which the intervention was a combination of individual and group counseling and included the women’s significant others [27]; however, as most included studies in the meta-analysis [17, 20, 22, 24, 25, 27], the Zhu study has an overall unclear risk of bias due to failures in blinding of personal and participants.

Uptake

Regarding the uptake outcome, the meta-analysis did not show a significant difference among the standard and enhanced counseling groups. However, study results were heterogeneous. Similar to what was described for the use outcome, in this case, the studies in which enhanced counseling did not seem to have a benefit showed a higher use incidence in the control group (Carneiro, 95.9%; Davidson, 95.8%; Langston, 91.7%), so it was not surprising that the intervention did not seem to be beneficial in these studies. However, the other two studies, which seem to show a benefit in the enhanced group, had a lower uptake incidence in the control group (Schunmann, 81.20%; Whitaker, 74.19%). This may be because their “uptake” definition did not include some hormonal contraceptive types, such as implant and monthly injectable.

LARC use has been the main objective of contraception campaigns in the last years [28]. We found a slight trend in enhanced counseling compared to standard counseling, which is represented by two studies with a larger population; however, in the meta-analyses, we found that “contraceptive use” is similar in the enhanced counseling and standard counseling groups. Of the three meta-analyzed studies for LARC use, while two suggested a benefit of the enhanced intervention, Bender 2004 [17] did not. However, given this study’s high risk of bias (including randomization problems causing the study groups to have different basal characteristics and lack of adjusted analysis for possible confounding factors), its results should be taken with caution. Currently, no clear consensus has been found that defines how long a person must be evaluated to guarantee the contraceptive method’s effectiveness. We consider that the evaluation period was noticeably short in the studies as the outcome was a long-term method.

Enhanced counseling at the time of abortion seems to reduce the occurrence of obstetric events; however, it is necessary to consider that the certainty of the evidence for all outcomes was very low, so future studies are needed to confirm this result and elicit if this possible benefit may be due to patient-centered counseling, explanation of myths, or other components [6] and if a higher effect can be achieved with a group or individual interventions in which the significant other is included [27].

No previous systematic reviews studied the difference between counseling provided by a physician and that by a non-physician. We did not conduct a meta-analysis for this comparison since the studies were quite different. Olavarrieta and Makenzius showed that, compared to physicians, non-physicians tend to prescribe more contraceptives but have fewer users taking contraceptives home [21, 23] and prescribe fewer IUDs [23]. However, more studies are needed to make a reliable conclusion.

Limitations and strengths

Some limitations must be considered when interpreting the results of this systematic review. Mainly, most RCTs did not show a minimally informative description of what counseling was provided to the control and intervention groups, which prevented them from fully understanding their results, which could be improved using TIDieR [12] and following the CONSORT checklist [29]. This is important since the effectiveness of counseling interventions may depend on several factors [30], including communication skills of the providers, considering key aspects of the patient’s life to develop the intervention [31, 32], and considering the woman’s family plan [6], need of information, and past experiences with contraception [7], in addition to their preferences that should be consistent with the contraceptive methods they use [33]. Likewise, the included studies did not report an evaluation of the counseling quality nor the user satisfaction.

Other important limitations were as follows: 1) few studies had a low risk of bias, and most failed in blinding. 2) Enhanced interventions were moderately heterogeneous across studies. It is expected to have certain heterogeneity since different contexts need different interventions. However, some counseling guidelines, such as those of the World Health Organization [20] and the United States Agency for International Development [34], can be used in future studies to establish certain components that are minimally assessed in contraceptive counseling interventions [34]. 3) Outcomes had different definitions across studies [7]. 4) Studies usually do not detail important information to understand their results, such as abortion restrictions and feasibility in their settings, although it seems that, of 11 studies, six were performed in countries where abortion is legal on request and three in a restrictive setting [7]. Future RCTs must consider these limitations.

Likewise, it is necessary to acknowledge that all included studies seem to have been performed considering sex or having a uterus and did not ask about gender identity, and we found no information regarding the effect of different types of counseling in specific groups, such as transgender man or nonbinary individuals, in which counseling may need to consider specific components [35].

However, to date, this is the most comprehensive systematic review that has assessed the effects of peri-abortion contraceptive counseling, summarizing information that may be useful to informed decision-making.