Development of the Medical eligibility criteria for contraceptive use, fifth edition

IDevelopment of the Medical eligibility criteria for contraceptive use, fifth edition

1.1. Background

1.1.1. Overview and scope of the guidelines

Over the past 40 years, there have been significant advances in the development of new contraceptive technologies, including changes in formulations and dosing, schedules for administration and novel delivery systems. However, current policies and health-care practices in some countries are based on scientific studies of contraceptive products that are no longer in wide use, on long-standing theoretical concerns that have never been substantiated or on the personal preference or bias of service providers. These outdated policies or practices often result in limitations to both the quality of, and the access to, family planning services for clients.

The goal of this document is to improve access to, and quality of, family planning services by providing policy-makers, decision-makers and the scientific community with recommendations that can be used for developing or revising national guidelines on medical eligibility criteria used in the provision of all hormonal contraceptives, intrauterine devices, barrier methods, fertility awareness-based methods, coitus interruptus, lactational amenorrhoea method, male and female sterilization, and emergency contraception. These evidence-based recommendations do not indicate a “best” method that should be used given a particular medical context; rather, review of the recommendations allows for consideration of multiple methods that could be used safely by people with certain health conditions (e.g. hypertension) or characteristics (e.g. age).

Because country situations and programme environments vary so greatly, it is inappropriate to set firm international guidelines on criteria for contraceptive use. However, it is expected that national programmes will use these recommendations for updating or developing their own contraceptive eligibility guidelines according to national health policies, needs, priorities and resources, while reflecting upon local values and preferences.

There are a total of four WHO guidance documents (cornerstones) pertaining to contraception; two that focus on evidence-based recommendations (primarily targeted towards policy-makers and programme managers) and two that focus on application of the recommendations (primarily targeted towards health-care providers). All four cornerstones are best interpreted and used in a broader context of reproductive and sexual health care. These four documents, listed below, are updated periodically to reflect changes in the medical and scientific knowledge.

Evidence-based recommendations for provision of contraception:

Medical eligibility criteria for contraceptive use (MEC) – provides guidance regarding “who” can use contraceptive methods safely; and
Selected practice recommendations for contraceptive use (SPR) – provides guidance regarding “how” to use contraceptive methods safely and effectively.

Practical tools for front-line providers of contraceptive counselling and services:

3.: Decision-making tool for family planning clients and providers – counselling tool that supports both provider and client in the process of choosing a contraceptive method; and

4.: Family planning: a global handbook for providers – offers evidence-based information on service delivery, method by method.

Figure 1The four cornerstones of family planning guidance

1.2. Methods

1.2.1. Development of earlier editions of the Medical eligibility criteria for contraceptive use

This document builds on a process initiated in 1994 to develop the first edition. The initial process involved comparing the eligibility criteria used by different agencies for various contraceptives, preparing summaries of published medical and epidemiological literature relevant to medical eligibility criteria, and preparing a draft classification for review by a larger group of experts and agencies. Two expert Working Group meetings were organized by WHO, in March 1994 and May 1995, to review the background classifications and to formulate recommendations; publication of the document followed in 1996.

Since the publication of the first edition of the MEC, the guideline has been revised and updated three times. With each revision, a Working Group of multidisciplinary experts was assembled to review newly published evidence pertaining to the topics addressed in the guideline. Moreover, with each revision, the Working Group used the opportunity to consider inclusion of new medical conditions and new contraceptive methods, as appropriate.

The second edition of the MEC was based on the recommendations of an expert Working Group meeting held at WHO on 8–10 March 2000, which brought together 32 participants from 17 countries, including representatives of many agencies and organizations. The Working Group reviewed new evidence since the last meetings in 1994 and 1995, primarily obtained from systematic reviews of the most recent literature.

The third edition of the MEC, was based on the recommendations of an expert Working Group meeting held at WHO on 21–24 October 2003, which gathered 36 participants from 18 countries, including representatives of many agencies and organizations. Systematic reviews of the evidence were prepared on topics with newly published evidence since the meeting in 2000; they were presented to the Working Group and provided the basis for their decision-making. A Guideline Steering Group (GSG), comprising seven external members, was established for this edition. The GSG was formed to advise WHO on behalf of the larger expert Working Group on matters related to emerging published evidence on topics covered by the guideline during interim periods between expert Working Group meetings.

The fourth edition of the MEC was based on the recommendations of an expert Working Group meeting held at WHO on 1–4 April 2008, which brought together 43 participants from 23 countries, including nine agency representatives. Eighty-six new recommendations were developed and 165 recommendations were revised for the fourth edition. All members of the expert Working Group were asked to declare any conflict of interest and three of the experts declared conflicts of interest relevant to the subject matter of the meeting. These conflicts of interest were determined not to be sufficient to preclude the experts from participating in the deliberations and development of recommendations and thus they were not asked to withdraw from this process.

The Guidelines Review Committee (GRC) was established by the Director-General of WHO in 2007 to ensure that WHO guidelines are of a high methodological quality and are developed through a transparent, evidence-based decision-making process. The fourth edition of the MEC was reviewed by the newly established GRC and was approved on 16 September 2009.

To assure that the guidelines remain current between guideline meetings, new evidence is identified through an ongoing comprehensive bibiliographic search (the Continuous Identification of Research Evidence, or CIRE system)¹. This evidence is synthesized and reviewed. In circumstances where new evidence warrants further evaluation, the GSG is tasked with evaluating such evidence and issuing interim guidance if necessary. Since the release of the fourth edition of the MEC, interim guidance has been issued twice. At the request of the GSG, WHO first convened a technical consultation on 26 January 2010 via teleconference to review new evidence regarding the risk of venous thromboembolism (VTE) in postpartum women. The teleconference brought together members of the GSG and three experts on VTE during the postpartum period. All participants in the consultation were asked to declare any conflict of interest; two participants declared a conflict of interest relevant to the subject matter, but they were not asked to withdraw from the process of recommendation formulation because the WHO Secretariat and GSG did not find these conflicts of interest sufficient to preclude them from participating in the deliberations and development of recommendations. The GRC approved the updated recommendations on 21 April 2010.

Following new findings of epidemiological studies regarding the use of hormonal contraception and HIV acquisition, progression and transmission, a second technical consultation was convened by WHO during 31 January – 1 February 2012. The meeting involved 75 individuals representing a wide range of stakeholders. Through a consensus-driven process, the group considered whether recommendations in the MEC pertaining to hormonal contraceptive use among women at high risk of HIV or women living with HIV should be changed in light of the accumulating evidence. All participants in the consultation were asked to declare any conflict of interest; 13 participants declared an academic conflict of interest relevant to the subject matter of the meeting. These conflicts of interest were determined not to be sufficient to preclude them from participating in the deliberations and development of recommendations and so they were not asked to withdraw from this process. The GRC approved the technical statement presenting the conclusions and updated recommendations of the meeting on 15 February 2012.

1.2.2. Development of the Medical eligibility for criteria for contraceptive use, fifth edition

In preparation for the fifth edition of the document, both approval for the planning and ultimately the final document were obtained from the GRC. Several key aspects of the updating process were adjusted to be in closer alignment with requirements set forth in the WHO handbook for guideline development, authored by the GRC Secretariat.² Specifically, these alterations included:

creation of groups with varying roles to undertake the revision;
convening an additional consultation to define the scope of the revision, giving priority to controversial topics and those for which new evidence had emerged, including topics addressed in interim guidance, clarifying recommendations with a Category 2/3 classification, and drafting questions relating to population, intervention, comparator and outcome (PICO questions) to guide the preparation of systematic reviews; and
applying the Grading Recommendations, Assessment, Development and Evaluation (GRADE) approach to evidence review and recommendation formulation.³

The groups responsible for the development of the fifth edition of the MEC included: a WHO Secretariat; an Evidence Secretariat including a GRADE methodologist; a Guideline Steering Group (GSG); and a Guideline Development Group (GDG), which was formerly called the expert Working Group for the earlier MEC editions. The GSG, which has served as an external advisory group to WHO on family planning guidelines since 2003, was part of the larger GDG, to be compliant with WHO requirements for guideline development and to gain input from a larger advisory group. For a summary of the members of the WHO Secretariat, the Evidence Secretariat and the GDG, see the Acknowledgements at the beginning of this document.

1.2.3. Prioritization of topics for the revision process

On 14–15 May 2013, the first GDG meeting convened in Ferney Voltaire, France, to initiate the revision process for the development of the fifth edition of the MEC. Prior to the meeting, the CIRE system¹ was used to identify recommendations from the fourth edition of the MEC for which new evidence was available.

To further inform decision-making with respect to clinical questions and priorities, the WHO Secretariat reached out to a broad group of stakeholders with expertise in family planning and familiarity with the guideline, including individuals from a number of implementing agencies, professional societies, and WHO regional and country offices, as well as the Ministry of Health in each of the Member States. They were asked to voluntarily complete an electronic 24-question anonymous survey available in English, French and Spanish, and to forward the link for the survey to others in their professional communities familiar with family planning and the MEC during the period 2 March – 2 May 2013. The respondents were asked to rank the importance of various outcomes pertaining to topics that had been identified as priority questions for the current revision, as well as to suggest other outcomes and clinical questions of importance, and to give input regarding the format of the guidance. More than 250 individuals submitted completed surveys; these results were presented to the GDG during the meeting to inform the prioritization process.

At the meeting, the WHO Secretariat presented brief summaries of new evidence to the GDG to determine whether the existing recommendation remained consistent or had become inconsistent with the updated body of evidence. Recommendations considered to be possibly inconsistent with the updated body of evidence were selected for presentation and discussion at a larger meeting convened in March 2014. Recommendations considered to be consistent with the updated body of evidence, and recommendations for which no new evidence had been identified through CIRE were determined by the GDG to need no further review during the revision process.

Also at this first GDG meeting, the members were asked to consider whether WHO should include several new conditions, contraceptive methods and/or formulations of methods, based upon their global relevance and availability in multiple countries. Participants were also asked to review the two interim guidance documents released since the fourth edition. Further, during this meeting the GDG was asked to address current recommendations which were classified as category “2/3” in the fourth edition, as earlier reviews by the GRC noted that these recommendations may be confusing to users of the document.

Thus, topics were prioritized for review and consideration by the GDG at the second meeting in March 2014 based on meeting one or more of the following criteria: topics identified as controversial or of particular importance to the field; topics with new evidence, for which the existing recommendation was potentially inconsistent with the updated body of evidence; topics with interim guidance issued by WHO since the MEC fourth edition; newly introduced contraceptive methods; or recommendations from the MEC fourth edition that were determined to lack clarity by the GRC. All existing recommendations that did not fall into one of these categories were reaffirmed by the GRC and thus were not reviewed.

For the topics outlined in Table 1.1, the GDG developed questions using the PICO format (i.e. questions with specified populations, interventions, comparators and outcomes) to serve as the framework for the systematic reviews and GRADE evidence tables. In order to inform the MEC recommendations, PICO questions generally guide the systematic review to focus on studies of populations with the condition or characteristic of interest using a specific contraceptive method compared with the same population not using the method, reporting on critical safety outcomes. PICO questions were also crafted to also identify relevant indirect evidence that may have included comparator populations without the condition or characteristic of interest using the same method, or reporting on surrogate outcomes. These systematic reviews, therefore, assessed the safety risks of using a given method among women with a particular medical condition or characteristic. The remainder of the existing recommendations were determined to be consistent with the body of published evidence and did not need to be formally reviewed for this revision.

Table 1.1

Medical eligibility criteria for contraceptive use, fifth edition: selection of topics for 2014 revision.

1.2.4. Evidence identification and synthesis

For each of the priority topics listed in Table 1.1, systematic reviews were conducted in accordance with PRISMA guidelines to answer PICO-formatted questions regarding safety outcomes.⁴ The systematic reviews may be accessed in Annex 2. In general, the PubMed and Cochrane databases were searched for studies published in any language in a peer-reviewed journal up to 15 January 2014, to inform the systematic reviews. Reference lists and direct contact with experts in the field were also used to identify other studies, including those in press; neither grey literature nor conference abstracts were included in these reviews. Due to heterogeneity of study designs, contraceptive formulations and outcome measures, meta-analyses were generally not performed. The quality of evidence presented in individual studies within a systematic review was assessed by review authors using the United States Preventive Services Task Force system.⁵ GRADE evidence profiles were then prepared by a GRADE methodologist to assess the quality of the summarized evidence and include the range of the estimates of effect for each clinical outcome assessed. GRADE evidence profiles were prepared for each PICO question for which evidence was found and clinical outcomes were reported. The systematic reviews that resulted from this process were peer-reviewed by selected members of the GDG, and final drafts were made electronically available to all GDG members prior to the consultations. Printed copies of GRADE evidence profiles for each topic were also given to each GDG member during the March 2014 GDG meeting. The written and orally presented systematic reviews and GRADE evidence profiles served as the basis for the GDG's deliberations.

1.2.5. Decision-making during the Guideline Development Group meetings

During 9–12 March 2014 and 24–25 September 2014, WHO convened a series of GDG meetings to review the evidence for the priority topics and, where appropriate, revise specific recommendations in the MEC. Members of the GDG and members of the External Peer Review Group (who did not participate in the GDG meeting) submitted Declaration of Interest forms to the WHO Secretariat: 14 individuals declared an academic conflict of interest relevant to the MEC guidance. The WHO Secretariat and the GDG reviewed all declarations of interest and, with the exception of two members (Dr Glasier and Dr Sitruk-Ware), found no conflicts of interest sufficient to preclude anyone from participating in the deliberations or development of recommendations. In the case of the two exceptions, the WHO Secretariat and the GDG agreed that their disclosed academic conflicts of interest were sufficient to preclude them from participating in the deliberations and development of recommendations relevant to ulipristal acetate (Dr Glasier) and the progesterone-releasing vaginal ring (Dr Sitruk-Ware). For details of the declared academic interests see Annex 1.

The GDG considered the overall quality of the safety evidence, paying particular attention to the strength and consistency of the data, according to the GRADE approach to evidence review. In most cases, the quality of evidence pertaining to each recommendation was low or very low and only addressed potential harms related to contraceptive use. To arrive at a category designation, within the range 1–4, the GDG considered these potential harms, the GRADE evidence profiles, the benefits of preventing unintended pregnancy, as well as the other GRADE constructs of values and preferences.

The GDG endorsed an approach to patient preferences and values that prioritized the availability of a wide range of contraceptive options, as women vary in their preferences regarding contraceptive selection and in the value they place on different beneficial and harmful outcomes.⁶ ⁷ In addition, the availability of a range of contraceptive options is critical because a woman's contraceptive choices are made at a particular time and in a particular societal and cultural context, and these choices are complex, multifactorial and subject to change.⁸ ⁹ Decision-making for contraceptive methods usually requires making trade-offs among the different methods, with advantages and disadvantages of specific contraceptive methods varying according to individual circumstances, perceptions and interpretations.

Owing to the focus of this guidance on the safety of specific contraceptive methods for women with medical conditions or personal characteristics, opportunity costs were not formally assessed during the formulation of these recommendations since costs may vary widely throughout different regions.¹⁰

Since publication of the first edition of the MEC in 1996, the 1–4 scale has been used to categorize medical eligibility for contraceptive use. These categories are well known by health-care providers, professional organizations, training institutions and ministries of health as the basis for determining contraceptive eligibility for women with medical conditions or characteristics. As a result, to avoid confusion and retain consistency, it was determined that recommendations would not be defined as “strong” or “weak” according to GRADE methodology and would instead retain the 1–4 scale reflecting eligibility for contraceptive use.

Through consensus, the GDG arrived at new and revised recommendations, as well as upholding the majority of the existing recommendations using the categories 1–4. For the topics they reviewed in 2014 (see Box 1.1), the GDG considered the potential benefits and risks of contraceptive method use with respect to each of the medical conditions or personal characteristics assessed.

Owing to the public health importance of recommendations on hormonal contraceptive use for women at risk of HIV and women living with HIV, and based on encouragement from the GDG, WHO issued its contraceptive eligibility guidance for women living with HIV or at high risk of acquiring the infection in advance of the entire guideline revision. The document, Hormonal contraceptive methods for women at high risk of HIV and living with HIV: 2014 guidance statement was approved by the WHO Guidelines Review Committee (GRC) on 7 July 2014. The statement was released on 24 July 2014, at the 20th International AIDS Conference.

A draft version of the entire MEC document was sent to the External Peer Review Group, comprising eight experts who did not participate in the GDG meeting. Comments received from these reviewers were addressed and incorporated into this guidance by the WHO Secretariat as appropriate. The final version of this document was approved by the GRC on 18 March 2015.

1.3. Dissemination and evaluation of the Medical eligibility criteria for contraceptive use, fifth edition

The recommendations in the Medical eligibility criteria for contraceptive use, fifth edition guidance were released during a global live Facebook Chat on 1 June 2015. A comprehensive dissemination and evaluation plan will be implemented, which will include widespread dissemination through the WHO regional and country offices, WHO Member States, the United Nations (UN) agency cosponsors of the Special Programme of Research, Development and Research Training in Human Reproduction (HRP) within the WHO Department of Reproductive Health and Research (i.e. UNDP, UNFPA, UNICEF, WHO and the World Bank), WHO collaborating centres, professional organizations, governmental and nongovernmental partner organizations working in the area of sexual and reproductive health, and civil society groups engaged in sexual and reproductive health projects The WHO Secretariat will work closely with sexual and reproductive health points of contact in the WHO regional offices to conduct a series of regional events during 2015–2016. In addition, special panel sessions will be organized during the summer and autumn of 2015 at international conferences convened by the International Society of Obstetricians and Gynaecologists (FIGO), the International Council of Nurses (ICN) and the International Confederation of Midwives (ICM) to update the membership of these societies about the revised recommendations. Once translations of the document in other official languages of the UN become available, opportunities to ensure effective dissemination will be actively sought. An evaluation survey targeting ministries of health, WHO offices and partners, professional organizations and civil society will be fielded to assess the extent and effectiveness of the dissemination, evaluate the level of implementation of the guidance into national policies, and identify areas for further refinement and research gaps in contraceptive eligibility criteria.

1.4. Reviewed recommendations

The Guideline Development Group (GDG) determined priority topics to be addressed as part of the revision process for the fifth edition (see Table 1.1).

Information on using the recommendations in practice, as well as recommendations in the fifth edition (new, revised and confirmed) are presented in Part II, sections 2.3 and 2.7, starting on p. 91. A summary of changes between the fourth edition of the MEC and the updated fifth edition is available in Part II (see section 2.6 and Tables 2.4–2.6, pp. 93–96).

1. Recommendations for combined hormonal contraceptives by age group

Question 1: Are women who use combined hormonal contraceptives (CHCs) at increased risk for fracture compared with women who do not use CHCs? (Direct evidence)

Study design	Randomized controlled trials, cohort studies and case-control studies
Population	Women of reproductive age
Intervention	CHC use
Comparator	Non-use of CHCs
Outcome	Fracture
Databases searched	PubMed and Cochrane Library

CHC use versus non-use of CHC; fracture risk (direct):	low
COC use versus non-use in adolescents; bone mineral density (indirect):	low
Combined contraceptive patch use versus non-use in adolescents; bone mineral density (indirect):	very low

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
Oral combined hormonal contraceptive use vs non-use
Fracture	3 cohort studies (n=128 255); 7 case-control studies (n=84 695 cases)	Serious limitations (1 good, 6 fair, 3 poor)	Serious inconsistency	No serious imprecision	Serious indirectness (no study specifically evaluated CHC use during adolescence)	No duration-response effect observed	Low	7 studies showed no association between CHC use and fracture risk, including 1 good-quality study (OR 1.05, 95% CI 0.86–1.29); 3 studies found a statistically significant but weak association (risk estimate range 1.07–1.3)^a

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
Oral combined hormonal contraceptive use vs non-use in adolescent women
Bone mass density (BMD)	1 RCT (n=83), 1 non-randomized trial (n=84); 11 cohort studies (n=3242)	Serious limitations (2 good, 5 fair, 6 poor)	Serious inconsistency	No serious imprecision	Serious indirectness (intermediate outcome)	1 study showed duration-response effect; variability in duration of follow-up	Low	9 studies showed oral CHC use associated with less BMD gain (or greater loss) than non-use, including the 2 good-quality studies; 4 studies showed no difference^a
Patch use vs non-use in adolescent women
BMD	1 non-randomized trial (n=10)	Very serious limitations (1 poor)	Cannot determine (1 study)	Very serious imprecision	Serious indirectness (intermediate outcome)	None	Very low	1 study found no effect of patch on BMD vs non-use

For COCs compared with progestogen-only pills (POPs), breastfeeding and infant outcomes:	low
For COCs compared with non-hormonal or non-use, breastfeeding continuation:	very low
For COCs compared with non-hormonal or non-use, breastfeeding duration:	very low
For COCs compared with non-hormonal or non-use, supplementation:	low
For COCs compared with non-hormonal or non-use, infant outcomes:	very low
For patch, ring, combined injectable contraceptives (CICs):	no evidence

For COCs, breastfeeding continuation:	low
For COCs, breastfeeding duration:	very low
For COCs, breastfeeding episodes:	very low
For COCs, supplementation:	low
For COCs, infant outcomes:	low
For patch, ring, CICs:	no evidence

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
Combined oral contraceptives (COCs) vs progestogen-only pills initiated at < 6 weeks postpartum
Breastfeeding continuation	1 RCT (n=127)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	Not applicable	Low	Ethinyl estradiol (EE) vs POP (1 RCT): 64% vs 64%, RR 1.0 (95% CI 0.78–1.3) at 8 weeks; 44% vs 41%, RR 1.1 (95% CI 0.71–1.6) at 6 months
Use of supplementation	1 RCT (n=127)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	Not applicable	Low	EE vs POP (1 RCT): no difference at 8 weeks (data not provided)
Infant growth	1 RCT (n=127)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	Not applicable	Low	EE vs POP (1 RCT): no difference in percent change in weight (P = 0.56), length (P = 0.41), or head circumference (P = 0.79) from weeks 2–8
COCs initiated at < 6 weeks postpartum vs non-hormonal or non-use
Breastfeeding continuation	1 RCT; 2 non-randomized studies (n=550)	Very serious limitations (1 fair-quality RCT; 2 poor-quality studies)	Serious inconsistency	No serious imprecision	Serious indirectness (older COC formulations with mestranol)	Variability in outcomes assessed and duration of follow-up	Very low	1 RCT found EE COC associated with lower likelihood of breastfeeding continuation vs placebo or Cu-IUD at 6 months (84% vs 91% vs 95%, RR 0.92 [95% CI 0.82–1.0] vs placebo and 0.88 [95% CI 0.79–0.97] vs IUD) but no difference at 12 months (61% vs 59% vs 65%); 1 poor-quality study of various COCs found no difference in breastfeeding continuation at 6 weeks; 1 poor-quality study found initiation of mestranol at 2 weeks associated with lower likelihood vs initiation at 6 weeks (RR 0.40, 95% CI 0.17–0.96) or non-use (RR 0.29, 95% CI 0.13–0.64)
Breastfeeding duration	1 cohort study (n=696)	Very serious limitations (1 poor)	Cannot determine (1 study)	No serious imprecision	Serious indirectness (older estrogen component of COC formulations)	None	Very low	Various COCs (quinestrol, EE or mestranol) vs no COC (1 study): 2.5 to 4.6 vs 5.3 months, P= 0.01 for quinestrol and mestranol vs no COC
Use of supplementation	2 RCTs (n=727)	Serious limitations (1 fair, 1 poor)	No serious inconsistency	Serious imprecision	Serious indirectness (largest study [n=451] evaluated mestranol COC)	Variability in duration of follow-up	Low	1 fair-quality RCT found EE COC associated with increased likelihood of supplementation vs injectable placebo (19.4% vs 7.8%, RR 2.5 [95% CI 1.1–5.4]) or oral placebo (19.4% vs 8.0%, RR 2.4 [95% CI 1.0–5.8]) at day 91; 1 poor-quality RCT found mestranol COC associated with increased likelihood (12.3% vs 3.4%, RR 3.6 [95% CI 1.7–8.1]) at day 8
Infant growth	3 RCTs (n=712); 3 observational studies (n=100)	Very serious limitations (1 fair-quality RCT; 5 poor-quality studies)	Serious inconsistency	No serious imprecision	Serious indirectness (most studies evaluated mestranol COC)	Variability in outcomes assessed and duration of follow-up	Very low	1 fair-quality RCT found EE COC associated with lower infant weight at 6 months (7864 vs 8333) and 1 year (9938 vs 10 746) vs placebo; 4 poor-quality studies of mestranol COC reported conflicting results vs placebo or no COC (1 study no difference, 1 study growth greater with COC, and 2 studies less weight gain with COC)

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
Combined oral contraceptive (COC) initiated at ≥ 6 weeks postpartum vs non-hormonal or non-use
Breastfeeding continuation	2 non-randomized studies (n=339)	Serious limitations (2 fair)	No serious inconsistency	No serious imprecision	No indirectness	None	Low	2 non-randomized studies found no difference between ethinyl estradiol (EE) COC vs non-COC (various) in breastfeeding continuation rates at 6 or 12 months (RR 0.96–0.99 in both studies)
Duration of breastfeeding	1 cohort study (n=96)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	1 cohort study found no difference between EE COC vs no oral contraceptives in duration of breastfeeding (3.7 vs 4.6 months, P < 0.05)
Breastfeeding episodes	1 cohort study (n=20)	Serious limitations (1 fair)	Cannot determine (1 study)	Very serious imprecision	No indirectness	None	Very low	1 small study found EE COC associated with more breastfeeding episodes than IUD on 7 of 21 days from postpartum day 42 to day 63 with no differences on other days
Use of supplementation	3 non-randomized studies; 1 cohort study (n=359 plus 1 study with < 50 women)	Serious limitations (3 fair, 1 poor)	Serious inconsistency	Serious imprecision	No indirectness	None	Low	2 fair-quality non-randomized studies found EE COC associated with increased likelihood of use of supplementation vs non-COC (various) at 3–12 months (differences ranged from 10–25% at various time points); 1 cohort study found no difference; 1 study found no difference between mestranol COC and IUD in age at supplementation
Infant growth	2 non-randomized studies; 3 cohort studies (n=455 plus 1 study with < 50 women)	Serious limitations (3 fair, 1 poor)	Serious inconsistency	Serious imprecision	No indirectness	None	Low	1 cohort study found EE COC associated with lower infant growth vs non-COC (various) from 3–4 months (599 vs 708 g) but not at 5–6 months and no difference in mean weight at 1 year; 1 non-randomized study found no difference in change in weight at 12–16 or 20–24 weeks; 1 non-randomized study found mestranol COC 0.1 mg with more infant growth than 0.075 mg or IUD; 1 cohort study found no difference in infant growth through 8 years; 1 cohort study showed no difference between EE COC and IUD from day 42 to 63

CHC use versus non-CHC use postpartum; VTE (direct):	very low
First 6 weeks postpartum versus non-pregnant, non-postpartum; VTE (indirect):	low

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
CHC use vs non-use in postpartum period
Venous thromboembolism (VTE)	1 cohort study (773 017 person-years)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	VTE (1 study): rate ratio 1.3 (95% CI 0.3–5.2) in weeks 0–6 for CHC use vs non-use; incidence 4.9 (95% CI 0.6–18) vs 3.5 (95% CI 3.1–3.9) per 1000 person-years in weeks 0–6 and 0.7 (95% CI 0.3–1.6) vs 0.3 (95% CI 0.2–0.5) per 1000 person-years in weeks 7–13

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
First 6 weeks postpartum vs non-pregnant, non-postpartum
Venous thromboembolism (VTE)	4 cohort studies (n=3 365 650); 1 case-control study (285 cases)	Serious limitations (1 good, 4 fair)	No serious inconsistency (direction consistent; variability in magnitude)	No serious imprecision	Serious indirectness (not women using CHC vs not using CHC)	Duration-response effect in first 6 weeks, first 1–3 weeks associated with highest risk;^a weeks 0–6 associated with higher risk than after week 7^b	Low	Any VTE (2 studies): rate ratio 21.5 (CI not available) and 22 (95% CI 18–27) and OR 84 (95% CI 32–223) and 12 (95% CI 7.9–18.6) DVT (1 study): incidence ratio 15 (95% CI 13–18) Pulmonary embolism (1 study): incidence ratio 9.2 (95% CI 6.5–13)

Use of CHCs versus non-use of CHCs; risk of VTE:	very low
Use of CHCs versus non-use of CHCs; risk of SVT:	very low

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
Use of oral contraceptives vs non-use in women with varicose veins
Deep vein thrombosis (DVT)	1 cohort study (96 335 women-years)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	IRR 5.97 in women with history of varicose veins; 4.42 in women with no history of varicose veins
Superficial venous thrombosis (SVT)	1 cohort (96 335 women-years)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	IRR 1.40 in women with history of varicose veins; 2.45 in women with no history of varicose veins

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
Use of oral contraceptives vs non-use in women with superficial venous thrombosis
Venous thromboembolism (VTE)	1 case-control study (1445 cases)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	OR 4.0 (95% CI 3.3–4.7) for OC use vs non-use with no history of SVT; OR 8.4 for OC use vs non-use with history of SVT (based on OR of 5.1 (95% CI 2.8–9.5) for no OC use and 43.0 (95% CI 15.5–119.3) OC use, reference no SVT history/no OC use)^a

Study design	Randomized controlled trials, cohort and case-control studies
Population	Women of reproductive age with dyslipidaemia
Intervention	CHC use
Comparator	Non-use of CHCs
Outcome	Arterial thrombotic events (e.g. myocardial infarction or thrombotic stroke), VTE or pancreatitis
Databases searched	PubMed and Cochrane Library

CHC use versus non-use of CHCs; ATE, VTE or pancreatitis (direct):	very low
Know dyslipidaemia versus no known dylipidaemia; ATE, VTE or pancreatitis (indirect):	very low
CHC use versus non-use of CHC; risk of lipid abnormalities (indirect):	very low

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
Use of combined oral contraceptives (COCs) vs non-use in women with hyperlipidaemia
Myocardial infarction	1 case-control study (248 cases)	Very serious limitations (1 poor)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	OR 7.5 for COC use vs no COC use, based on OR 24.7 (95% CI 5.6–108.5) for COC use/hyperlipidemia and 3.3 (95% CI 1.6–6.8) for no COC use/hyperlipidaemia, reference no COC use/no hyperlipidaemia

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
Women with hyperlipidaemia vs no hyperlipidaemia prescribed combined oral contraceptives
Deep vein thrombosis (DVT) and pulmonary embolism (PE)	1 cohort study (n=329 995)	Very serious limitations (1 poor)	Cannot determine (1 study)	No serious imprecision	Serious indirectness (not COC use vs non-use)	None	Very low	8.51 vs 6.14 per 10 000 woman-years, IRR 1.39 (95% CI 1.04–1.85)

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
Worsening hyperlipidemia in users of combined oral contraceptives vs non-users
Transient ischemic attacks and cerebrovascular accidents	1 cohort study (n=329 995)	Very serious limitations (1 poor)	Cannot determine (1 study)	No serious imprecision	Serious indirectness (not COC use vs non-use)	None	Very low	10.14 vs 5.76 per 10 000 woman years, IRR 1.76 (95% CI 1.51–2.06)

Breastfeeding continuation, breastfeeding duration:	low
Use of supplementation:	very low

IDevelopment of the Medical eligibility criteria for contraceptive use, fifth edition

1.1. Background

1.1.1. Overview and scope of the guidelines

Figure 1The four cornerstones of family planning guidance

1.2. Methods

1.2.1. Development of earlier editions of the Medical eligibility criteria for contraceptive use

1.2.2. Development of the Medical eligibility for criteria for contraceptive use, fifth edition

1.2.3. Prioritization of topics for the revision process

1.2.4. Evidence identification and synthesis

1.2.5. Decision-making during the Guideline Development Group meetings

1.3. Dissemination and evaluation of the Medical eligibility criteria for contraceptive use, fifth edition

1.4. Reviewed recommendations

1. Recommendations for combined hormonal contraceptives by age group

Recommendations

Remarks

Summary of the evidence

Quality of the evidence

References

2. Recommendations for combined hormonal contraceptives among breastfeeding women

Recommendations

Remarks

Summary of the evidence

Quality of the evidence

References

3. Recommendations for combined hormonal contraceptives among postpartum women

Recommendations

Remarks

Summary of the evidence

Quality of the evidence

References

4. Recommendations for combined hormonal contraceptives among women with superficial venous disorders

Recommendations

Remarks

Summary of the evidence

Quality of the evidence

References

5. Recommendations for combined hormonal contraceptives among women with dyslipidaemias

Recommendations

Remarks

Summary of the evidence

Quality of the evidence

References

6. Recommendations for progestogen-only contraceptives and levonorgestrel-releasing intrauterine devices among breastfeeding women

6a. Recommendations for use of progestogen-only contraceptives (pills, injectables, implants)

< 6 weeks postpartum

≥ 6 weeks to < 6 months postpartum

≥ 6 months postpartum

6b. Recommendations for use of levonorgestrel-releasing intrauterine devices (LNG-IUDs)

< 48 hours postpartum

≥ 48 hours to < 4 weeks postpartum

≥ 4 weeks postpartum

Puerperal sepsis

Remarks

Summary of the evidence

Quality of the evidence

Breastfeeding outcomes

Infant outcomes

Breastfeeding outcomes

Infant outcomes

References

7. Recommendations for safety of depot medroxyprogesterone acetate delivered subcutaneously

Recommendations

Age

Endometriosis

HIV

Obesity

Remarks

Summary of the evidence

Quality of the evidence

References

8. Recommendations for safety of Sino-implant (II)

Recommendations

Remarks

Summary of the evidence

Quality of the evidence

References

9. Recommendations for use of emergency contraceptive pills, including adding the condition of obesity and the new method, ulipristal acetate

Recommendations

Remarks

Summary of the evidence

Breastfeeding duration and use of supplementation:	low
Breastfeeding continuation:	very low

Breastfeeding duration:	low
Breastfeeding continuation and use of supplementation:	very low

Factor	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
Progestogen-only pill (POP) vs combined oral contraceptive (COC) initiated at < 6 weeks postpartum
Breastfeeding continuation	1 RCT (n=127)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	NA	Low	Norethindrone vs ethinyl estradiol (EE) COC (1 RCT): 64% vs 64%, RR 0.99 (95% CI 0.76–1.3) at 8 weeks; 41% vs 44%, RR 0.94 (95% CI 0.63–1.4) at 6 months
Use of supplementation	1 RCT (n=127)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	NA	Low	Norethindrone vs EE COC (1 RCT): no difference at 8 weeks (data not provided)
Infant growth	1 RCT (n=127)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	NA	Low	Norethindrone vs EE COC (1 RCT): no difference in percent change in weight (P = 0.56), length (P = 0.41), or head circumference (P = 0.79) from weeks 2–8
LNG-IUD initiated at < 6 weeks postpartum vs non-hormonal contraception
Breastfeeding continuation	1 RCT (n=110)	Very serious limitations (1 poor)	Cannot determine (1 study)	Serious imprecision	No indirectness	NA	Very low	LNG-UD (30 mcg/d or 10 mcg/d) vs Cu-IUD (1 RCT): 58% vs 79% at 8 months, RR 0.74 (95% CI 0.57–0.95)
Breastfeeding duration	1 RCT (n=110)	Very serious limitations (1 poor)	Cannot determine (1 study)	Serious imprecision	No indirectness	NA	Very low	LNG-IUD 30 mcg/d vs LNG IUD 10 mcg/d vs Cu-IUD (1 RCT): 197 vs 182 vs 208 days (P > 0.05)
Infant growth	1 RCT (n=110)	Very serious limitations (1 poor)	Cannot determine (1 study)	Serious imprecision	No indirectness	NA	Very low	LNG-IUD (30 mcg/d or 10 mcg/d) vs Cu-IUD (1 RCT): no differences through 12 months
Progestogen-only injectable initiated at < 6 weeks postpartum vs non-hormonal
Breastfeeding continuation	3 cohort studies (n=617)	Very serious limitations (1 fair, 2 poor)	Cannot determine (1 study)	Serious imprecision	No indirectness	Variability in interventions and outcomes^a	Very low	No clear differences in 2 studies; in a 3rd study weaning occurred later with DMPA or NET-EN
Use of supplementation	5 cohort studies (n=1370)	Very serious limitations (1 fair, 4 poor)	No serious inconsistency	No serious imprecision	No indirectness	Variability in interventions and outcomes^a	Low	All 5 studies found DMPA or NET-EN associated with similar or lower likelihood of exclusive breastfeeding
Duration of breastfeeding	5 cohort studies (n=1732)	Very serious limitations (1 fair, 4 poor)	No serious inconsistency	No serious imprecision	No indirectness	None	Low	All 5 studies found DMPA associated with no difference or increased duration of breastfeeding vs non-hormonal methods
Infant growth	6 cohort studies (n=4403)	Very serious limitations (5 poor)	No serious inconsistency	No serious imprecision	No indirectness	None	Low	None of 5 studies found DMPA or NET-EN associated with decreased infant growth; 1 study found progestogen-only injectable associated with increased weight gain through 3 months
POPs initiated at < 6 weeks postpartum vs non-hormonal
Breastfeeding continuation	1 non-randomized trial (n=273) 3 cohort studies (n=756)	Serious limitations (1 fair-quality cohort study, 3 poor-quality studies)	No serious inconsistency	No serious imprecision	No indirectness	Variability in interventions and outcomes^a	Low	None of 4 studies found various POPs associated with lower likelihood of breastfeeding continuation; 2 studies found POPs associated with higher likelihood of breastfeeding continuation
Breastfeeding initiation	1 RCT (n=20) and 1 nonrandomized trial (n=20)	Very serious limitations (2 poor)	No serious inconsistency	Very serious imprecision	No indirectness	Variability in interventions and outcomes^a	Very low	1 RCT found no difference between norethisterone at ≤ 14 hours postpartum (PP) vs placebo in breastfeeding initiation; 1 non-randomized trial found lynestrenol at 2 days PP associated with initiation of breastfeeding at 3 vs 5 days PP with placebo
Duration of breastfeeding	2 cohort studies (n=572)	Very serious limitations (2 poor)	No serious inconsistency	Serious imprecision	No indirectness	Variability in interventions and outcomes^a	Low	2 studies found POPs associated with somewhat longer duration of breastfeeding vs non-hormonal comparators
Use of supplementation	2 cohort studies (n=1000)	Very serious limitations (1 fair, 1 poor)	Serious inconsistency	No serious imprecision	No indirectness	Variability in interventions and outcomes^a	Very low	1 fair-quality study found norgestrel associated with more frequent supplementary feeding but no difference in proportion of women supplementing; 1 poor-quality study found LNG associated with somewhat later initiation of supplementation (5.4 vs 4.6 months PP)
Infant growth	1 RCT (n=20), 1 non-randomized trial (n=20); 3 cohort studies (n=1083)	Serious limitations (2 fair-quality cohort studies, 3 poor-quality studies)	No serious inconsistency	No serious imprecision	No indirectness	Variability in interventions and outcomes^a	Very low	4 studies found no difference on measures of infant growth; 1 small, poor-quality, non-randomized study found greater increase with lynestrenol than placebo
Progestogen-containing implants initiated at < 6 weeks postpartum vs non-hormonal
Breastfeeding continuation	3 cohort studies (n=520)	Very serious limitations (1 fair, 2 poor)	Serious inconsistency	Serious imprecision	No indirectness	Variability in interventions and outcomes^a	Very low	3 studies reported conflicting findings regarding effects of progestogen-containing implants on measures of breastfeeding continuation; the 1 fair-quality study found no difference between nomegestrol implant in 2nd month PP vs Cu-IUD
Breastfeeding episodes	2 cohort studies (n=392)	Very serious limitations (1 fair, 1 poor)	No serious inconsistency	Serious imprecision	No indirectness	Variability in interventions and outcomes^a	Very low	2 studies found no differences in breastfeeding frequency
Duration of breastfeeding	1 cohort study (n=80)	Serious limitations (1 fair)	Cannot determine (1 study)	Very serious imprecision	No indirectness	None	Very low	1 fair-quality cohort study found no difference between etonogestrel (ETG) implant at 28–56 days PP vs Cu-IUD in duration of breastfeeding
Use of supplementation	3 cohort studies (n=430)	Very serious limitations (1 fair, 2 poor)	Serious inconsistency	Serious imprecision	No indirectness	Variability in interventions and outcomes^a	Very low	2 studies found no difference in use of supplementation; 1 study found norethindrone associated with increased likelihood of supplementation at 3 months
Infant growth	6 cohort studies (n=870)	Serious limitations (2 fair, 4 poor)	No serious inconsistency	No serious imprecision	No indirectness	Variability in interventions and assessment of outcomes^a	Low	2 fair-quality and 2 poor-quality studies found no difference in measures of infant growth; 1 poor-quality study found LNG associated with more weight gain than Cu-IUD; 1 poor-quality study found LNG associated with slower weight gain than Cu-IUD or barrier/no method
Multiple progestogen-only methods initiated at < 6 weeks postpartum vs non-hormonal
Breastfeeding continuation	1 cohort study (n=319)	Very serious limitations (1 poor)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	1 study found no difference between LNG implant or POP prior to discharge vs non-hormonal contraception in breastfeeding continuation at 2–6 weeks PP
Use of supplementation	1 cohort study (n=319)	Very serious limitations (1 poor)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	1 study found no difference between LNG implant or POP prior to discharge vs non-hormonal contraception in use of supplementation at 2–6 weeks PP
Non-orally available progesterone initiated at < 6 weeks postpartum vs non-hormonal
Breastfeeding continuation	2 cohort studies (n=1092)	Very serious limitations (1 fair, 1 poor)	Serious inconsistency	Serious imprecision	No indirectness	None	Very low	1 fair-quality study found progesterone pellets at 30–35 days PP associated with decreased likelihood of breastfeeding at 6 months (51% vs 58%) and 12 months (11% vs 18%); 1 poor-quality study found no difference between progesterone pellets at 30 or 60 days vs Cu-IUD or placebo injection in breastfeeding rates
Use of supplementation	2 cohort studies (n=1092)	Very serious limitations (1 fair, 1 poor)	No serious inconsistency	Serious imprecision	No indirectness	None	Very low	2 studies found no differences in use of supplementation
Infant growth	2 cohort studies (n=1092)	Very serious limitations (1 fair, 1 poor)	No serious inconsistency	Serious imprecision	No indirectness	None	Very low	2 studies found no differences in infant growth
LNG-IUD initiated at > 6 weeks postpartum vs non-hormonal
Duration of breastfeeding	1 RCT (n=320)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	LNG-IUD at 6–8 weeks PP vs Cu-IUD (1 RCT): 149 vs 160 days
Use of supplementation	1 RCT (n=320)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	LNG-IUD at 6–8 weeks PP vs Cu-IUD (1 RCT): no difference in exclusive breastfeeding
Infant growth	1 RCT (n=320)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	LNG-IUD at 6–8 weeks PP vs Cu-IUD (1 RCT): no difference in infant growth
Progestogen-only injectable initiated at > 6 weeks postpartum vs non-hormonal
Breastfeeding continuation	1 RCT (n=170)	Very serious limitations (1 poor)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	DMPA at 6 weeks PP vs non-hormonal (1 RCT): no difference in rates of discontinuation
Duration of breastfeeding	1 cohort study (n=1538)	Serious limitations (1 fair)	Cannot determine (1 study)	No serious imprecision	No indirectness	None	Low	DMPA or NET-EN at 6–8 weeks PP vs non-hormonal contraception (1 cohort study): no difference in duration of breastfeeding
Use of supplementation	1 RCT (n=170); 1 cohort study (n=212)	Very serious limitations (2 poor)	No serious inconsistency	Serious imprecision	No indirectness	None	Very low	2 studies found no differences in use of supplementation
Infant growth	1 RCT (n=170); 2 cohort studies (n=1750)	Very serious limitations (1 fair-quality cohort study, 2 poor-quality studies)	No serious inconsistency	No serious imprecision	No indirectness	Variability in outcomes^a	Low	None of 3 studies found decreased infant growth with DMPA or NET-EN; 2 studies reported some findings suggesting greater weight gain
POP initiated at > 6 weeks postpartum vs non-hormonal
Breastfeeding continuation	1 RCT (n=144)	Very serious limitations (1 poor)	Cannot determine (1 study)	Very serious imprecision	No indirectness	None	Very low	Norgestrel at 6 weeks PP vs non-hormonal contraception (1 RCT): no difference in discontinuation of breastfeeding
Duration of breastfeeding	2 cohort studies (n=1709)	Serious limitations (2 fair)	No serious inconsistency	No serious imprecision	No indirectness	Variability in interventions^a	Low	2 studies found no difference in breastfeeding duration
Use of supplementation	1 RCT (n=144); 1 nonrandomized trial (n=120)	Very serious limitations (2 poor)	Serious inconsistency	No serious imprecision	No indirectness	Variability in interventions and outcomes^a	Very low	1 RCT found no difference in use of supplementation and 1 non-randomized trial found lower mean age at supplementation with lynestrenol vs IUD + placebo (11 vs 15 weeks, P not reported)
Infant growth	1 RCT (n=144); 3 nonrandomized studies (n=1829)	Serious limitations (2 fair-quality cohort studies, 2 poor-quality studies)	No serious inconsistency	No serious imprecision	No indirectness	Variability in interventions and outcomes^a	Low	4 studies found no difference in measures of infant growth
Progestogen-only implant or progestogen-containing IUD initiated at > 6 weeks postpartum vs non-hormonal
Breastfeeding continuation	2 cohort studies (n=57)	Very serious limitations (2 poor)	No serious inconsistency	Very serious imprecision	No indirectness	None	Very low	2 studies found no difference in breastfeeding rates
Breastfeeding duration	4 cohort studies (n=2329)	Serious limitations (4 fair)	No serious inconsistency	No serious imprecision	No indirectness	Variability in outcomes^a	Low	4 studies found no difference in breastfeeding duration
Use of supplementation	3 cohort studies (n=549)	Serious limitations (2 fair, 1 poor)	No serious inconsistency	Serious imprecision	No indirectness	Variability in interventions and outcomes^a	Low	3 studies found no difference in use of supplementation
Infant growth	6 cohort studies (n=2386)	Serious limitations (4 fair, 2 poor)	No serious inconsistency	No serious imprecision	No indirectness	Variability in interventions and outcomes^a	Low	6 studies found no difference in measures of infant growth
Multiple progestogen-only methods initiated at > 6 weeks postpartum vs non-hormonal
Breastfeeding duration	1 cohort study (n=34)	Very serious limitations (1 poor)	Cannot determine (1 study)	Very serious imprecision	No indirectness	None	Very low	DMPA, POP or LNG-IUD vs non-hormonal contraception (1 cohort study): 183 vs 183 days (P =0.38)
Infant growth	1 cohort study (n=140)	Very serious limitations (1 poor)	Cannot determine (1 study)	Very serious imprecision	No indirectness	None	Very low	DMPA at 6 weeks PP vs non-hormonal contraception (1 cohort study): no difference in weight through 26 weeks PP
Non-orally available progesterone initiated at > 6 weeks postpartum vs non-hormonal
Breastfeeding continuation	2 cohort studies (n=788)	Very serious limitations (1 fair, 1 poor)	No serious inconsistency	No serious imprecision	No indirectness	Variability in interventions^a	Very low	1 fair-quality study found elcometrine implant associated with higher breastfeeding rate at 3 and 5 months (but not at 9 and 12 months); 1 poor-quality study found no difference between progesterone pellets vs Cu-IUD or placebo at 6 or 13 months PP
Breastfeeding duration	1 cohort study (n=200)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	Nesterone implant at 55–60 days PP vs Cu-IUD (1 cohort study): 273 vs 263 days (NS)
Infant growth	3 cohort studies (n=988)	Serious limitations (2 fair, 1 poor)	No serious inconsistency	No serious imprecision	No indirectness	Variabilty in interventions and outcomes^a	Low	3 studies found no differences in measures of infant growth
Earlier vs later initiation of progestogen-only methods
Breastfeeding continuation	2 RCTs (n=165) and 1 cohort study (n=35)	Serious limitations (2 fair-quality RCTs, 1 poor-quality cohort study)	No serious inconsistency	Serious imprecision	No indirectness	Variabilty in interventions and outcomes^a	Low	1 RCT found immediate LNG-IUD associated with lower breastfeeding rate at 6 months vs initiation at 6–8 weeks PP (6% vs 24%, P = 0.02); 1 cohort study found norethindrone implant at 6 days PP associated with lower rate at 8 months (57% vs 67%); 1 RCT found no difference in rate of lactation failure
Breastfeeding duration	1 RCT (n=96)	Serious limitations (1 fair)	Not applicable 1 study)	Very serious imprecision	No indirectness	None	Very low	Immediate LNG-IUD vs 6–8 weeks PP (1 RCT): 5 vs 8.5 weeks (P=0.06)
Use of supplementation	3 RCTs (n=205) and 4 cohort studies (n=660)	Serious limitations (3 fair-quality RCTs, 1 fair- and 3 poor-quality cohort studies)	Serious inconsistency	No serious imprecision	No indirectness	Variability in interventions and outcomes^a	Very low	Inconsistent effects on use of supplementation among 7 studies
Infant growth	1 RCT (n=40) and 3 cohort studies (n=543)	Very serious limitations (1 fair-quality RCT, 3 poor-quality cohort studies)	No serious inconsistency	Serious imprecision	No indirectness	Variability in interventions and outcomes^a	Low	4 studies found no differences in measures of infant growth

DMPA-SC and age:	very low
DMPA-SC and endometriosis:	very low
DMPA-SC and HIV:	very low
DMPA-SC and obesity:	very low
DMPA-SC versus DMPA; contraceptive efficacy (indirect):	very low
DMPA-SC versus DMPA; weight gain (indirect):	very low
DMPA-SC versus DMPA; changes in bleeding pattern (indirect):	very low.

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
DMPA-SC use in women in different BMI categories
Contraceptive efficacy	1 RCT and 1 cohort from an RCT (n=2321)	Serious limitations (2 fair)	No serious inconsistency	No serious imprecision	No indirectness	None	Very low	No pregnancies in women in all BMI categories
Weight gain	2 non-randomized studies (n=2336)	Serious limitations (1 fair-quality and 1 poor)	No serious inconsistency	No serious imprecision	No indirectness	None	Very low	No differences in weight gain across BMI categories
Changes in bleeding patterns	1 study with pooled data from 3 studies (n=2321)	Serious limitations (1 fair)	Cannot determine (1 study)	No serious imprecision	No indirectness	None	Very low	No differences in bleeding patterns across BMI categories
Change in bone mineral density	1 cohort study (n=15)	Very serious limitations (1 poor)	Cannot determine (1 study)	Very serious imprecision	No indirectness	None	Very low	No differences in BMD changes across BMI categories
DMPA-SC use in women in different age groups
Weight gain	1 study with pooled data from 3 studies (n=2321)	Serious limitations (1 fair)	Cannot determine (1 study)	No serious imprecision	No indirectness	None	Very low	No differences in weight gain across age categories
Changes in bleeding patterns	1 study with pooled data from 3 studies (n=2321)	Serious limitations (1 fair)	Cannot determine (1 study)	No serious imprecision	No indirectness	None	Very low	No differences in bleeding patterns across age categories
Changes in bone mineral density	1 RCT (n=534)	No serious limitations (1 good)	Cannot determine (1 study)	Serious imprecision	No indirectness	BMD changes not reported by age category, enrolled women aged 18–35 years	Very low	Smaller decrease in median BMD with DMPA-SC vs DMPA-IM
DMPA-SC in women with endometriosis
Weight gain	2 uncontrolled studies (n=289)	Very serious limitations (1 fair)	No serious inconsistency	Serious imprecision	No indirectness	No comparison group of women without endometriosis	Very low	Weight gain 0.70–0.95 kg at 6 months and 0.90–1.35 kg at 18 months
Change in bleeding pattern	2 uncontrolled studies (n=289)	Very serious limitations (1 fair)	No serious inconsistency	Serious imprecision	No indirectness	No comparison group of women without endometriosis	Very low	Increase in amenorrhoea 20% at 3 months and 24% at 6 months in 1 study; 1 study reported 35 bleeding or spotting days in 1st 90 days and 24 in 2nd 90 days
DMPA-SC vs DMPA-IM in women with HIV
Contraceptive efficacy	1 RCT (n=357)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	Pregnancy rate 1.1% with both DMPA-SC and DMPA-IM
Change in bleeding pattern	1 RCT (n=357)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	No differences in menstrual irregularity, heavier bleeding

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
DMPA-SC vs DMPA-IM use in healthy women
Contraceptive efficacy	1 RCT (n=534)	No serious limitations (1 good)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	Low rates of pregnancy with DMPA-SC and DMPA-IM
Weight gain	1 study with pooled data from 3 studies (n=2321)	Serious limitations (1 fair)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	Median weight gain 4.5 kg with DMPA-SC vs 5.8 kg with DMPA-IM at 36 months
Change in bleeding pattern	1 RCT (n=534)	No serious limitations (1 good)	Cannot determine (1 study)	Serious imprecision	No indirectness	None	Very low	No differences in rates of intermenstrual bleeding or amenorrhoea between DMPA-SC vs DMPA-IM

Sino-impant (II) versus non-use of Sino-implant (II); various outcomes (indirect):
Ectopic pregnancy:	low
Weight gain:	moderate
Blood loss:	low
Bone mineral density:	very low
Blood pressure:	low
Other adverse events:	very low
Pregnancy:	very low

Outcome	Type and number of studies (number of participants)	Limitations	Inconsistency	Imprecision	Indirectness	Other factors	Quality	Estimate of effect
Sino-implant (II) vs Sino-implant (I) or Norplant in healthy women
Ectopic pregnancy	4 RCTs (24 972)	Serious limitations (4 fair)	No serious inconsistency	Serious imprecision	No indirectness	None	Low	Very few ectopic pregnancies and no clear difference in risk in 4 RCTs of SI (II) vs SI (I) or Norplant
Weight gain	3 RCTs (n=4443); 1 cohort study (n=617)	Serious limitations (2 fair- and 1 poor-quality RCT, 1 very poor-quality cohort study)	No serious inconsistency	No serious imprecision	No indirectness	None	Moderate	No difference between SI (II) and SI (I) or Norplant in weight gain in 3 studies; less weight gain with SI (II) than control (no method or non-hormonal) in 1 study
Blood loss, change in haemoglobin	2 RCTs (n=389)	Serious limitations (1 fair, 1 poor)	No serious inconsistency	Serious imprecision	No indirectness	None	Low	No clear difference in blood loss or haemoglobin levels with SI (II) vs SI (I) or Norplant
Bone mineral density (BMD)	1 cross-sectional study (n=166)	Very serious limitations (1 poor)	Cannot determine (1 study)	Very serious imprecision	Serious indirectness (intermediate outcome)	None	Very low	No difference between SI (II), SI (I) and Norplant in BMD after ≥ 3 years
Blood pressure (BP)	1 RCT (n=2297); 1 cohort study (n=617)	Very serious limitations (1 fair-quality RCT, 1 very poor-quality cohort study)	No serious inconsistency	No serious imprecision	No indirectness	Variability in measures of blood pressure effects	Low	1 RCT found very few cases of increased BP with SI (II), SI (I), or Norplant; 1 cohort study found higher BP with control (no method or non-hormonal) than SI (II)
Other adverse events	2 RCTs (n=22 672)	Serious limitations (2 fair)	No serious inconsistency	No serious imprecision	No indirectness	Variability in outcomes measured	Very low	No difference between SI (II) and SI (I) or Norplant in various adverse events or reasons for removal
Effects of weight on contraceptive efficacy in women using Sino-implant (II)
Pregnancy	2 RCTs (n=10 940)	Serious limitations (2 fair)	Serious inconsistency	Serious imprecision	No indirectness	None	Very low	1 RCT found higher pregnancy rate in women ≥ 70 kg than women < 60 kg (P not reported); 1 RCT found no association between weight and risk of pregnancy

Study design	Primary research articles in all languages, including pharmacokinetic studies
Population	Women with characteristics or medical conditions outlined in the Medical eligibility criteria (MEC) update (pregnancy, breastfeeding, past ectopic pregnancy, history of severe cardiovascular complications, angina pectoris, migraine, severe liver disease, CYP3A4 inducers, repeated [ECP] use and rape).
Intervention	Use of hormonal ECPs (COCs, LNG or UPA)
Comparator	Non-use of hormonal ECPs
Outcome	Any adverse events (did not include side-effects)
Databases searched	PubMed and Cochrane Library

LNG-ECP use versus non-use of LNG-ECP; breastfeeding outcomes:	very low
LNG-ECP use versus non-use of LNG-ECP; infant growth/behaviour:	very low