Beyond Early Attrition: Access Barriers as the Leading Documented Reason for Stopping Monthly Norethisterone Enanthate/Estradiol Valerate Among Established Users in China: A Retrospective Cohort Study

Abstract

Purpose

Most women who begin monthly norethisterone enanthate 50 mg/estradiol valerate 5 mg (NET-EN/E2V) after an abortion stop within a year; less is known about those who continue. We describe use status, duration, menstrual change and documented reasons for stopping among established users in China.

Patients and Methods

We reviewed routine records of 134 women who had received at least six NET-EN/E2V injections at a tertiary maternal and child health hospital in Wuxi, Jiangsu Province, China. Women started between May 2023 and November 2025, with abstraction through May 2026. Duration analyses covered the 98 women no longer receiving injections at data cut-off; reason analyses covered the 61 with a documented reason. We compared post-abortion with other initiation contexts and fitted an exploratory logistic regression for use of at least 12 months.

Results

At data cut-off, 36 of 134 women (26.9%) were still receiving injections. Among the 98 no longer active, mean completed-use duration was 11.7±5.8 months and 40 (40.8%; 95% CI 31.6–50.7) reached at least 12 months. Initiation context predicted neither active status (P=0.418) nor duration (P=0.713). Among the 61 women with a documented reason, practical access barriers accounted for 30 (49.2%; 95% CI 37.1–61.4), menstrual change for 18 (29.5%), other side effects for 10 (16.4%) and contraceptive failure for three (4.9%). Practical barriers accounted for 43.5% at 6–11 months and 66.7% at ≥12 months.

Conclusion

Among established users with a documented reason for stopping, practical access barriers were recorded more often than pharmacological intolerance, and their share was larger after longer use. Because a reason was documented for only 61 of 98 women who stopped, and earlier discontinuers were excluded by design, these proportions describe documented reasons and do not estimate continuation from initiation. Easing the scheduling burden of monthly clinic visits is an actionable service target.

Plain Language Summary

Women who choose a monthly contraceptive injection need to visit a clinic every month for the next dose. A recent study in Shanghai found that most women who chose this method after an abortion stopped within a year. We wanted to learn about a different group: the women who kept going past the first six injections.

We read the routine clinic records of 134 women in China. At the end of the study period, 36 women were still coming for their injections. Of the 98 women who had stopped, 40 had used the method for a year or longer and six for two years or longer. The records gave a reason for stopping for 61 women. For 30 of them the reason was practical: the clinic visit was hard to fit in, an appointment slipped by, or they moved to another city. For 18 the reason was a change in their periods, for 10 another side effect, and for three a pregnancy.

Among the women whose records gave a reason, practical access problems were recorded most often. They made up a larger share of the reasons documented among women with 12 months of use or more, although only 15 women were in that group. Services could address these problems by sending appointment reminders, offering more flexible visit times, and helping women continue injections after moving to another clinic. Because we studied women who had already received six injections, our findings do not describe everyone who starts the method.

Introduction

The real-world effectiveness of any user-dependent contraceptive method depends on adherence to the dosing schedule. Combined injectable contraceptives (CICs) contain an estrogen and a progestogen and are given as a single intramuscular injection each month. Norethisterone enanthate 50 mg combined with estradiol valerate 5 mg (NET-EN/E2V) is one of the most widely used CIC formulations and suits medically eligible people who prefer a monthly, provider-administered method that requires neither a daily action nor a procedure.^1,2

Phase III trials and systematic reviews have established the efficacy, cycle control, safety, and reversibility of NET-EN/E2V.^3–6 These studies also showed that sustained use is achievable under trial conditions: in a multicenter Latin American Phase III program, continuation reached 64.7% at 12 months and 31.0% at 24 months, and cumulative discontinuation attributable to bleeding problems was 6.1% at 36 months.⁵ Trial participants, however, are screened, counseled, and actively followed, and injections are typically provided free of charge. Routine services differ on each of these points, so trial continuation cannot be carried over to routine care.

Two distinct processes can end an episode of CIC use. The first is pharmacological: bleeding change, another side effect, or a safety concern makes the method unacceptable. The second is logistical: the woman still wants the method but cannot reliably obtain the next dose. A monthly injectable is unusual in that it renews the second risk every four weeks, whereas an intrauterine device or an implant requires no return visit once placed. Separating the two processes matters because they call for different responses — counseling, method switching, or clinical management in the first case, and changes to scheduling, reminders, and continuity of supply in the second. Routine-care records can support this distinction without any assumption about individual motivation.

In China, post-abortion contraceptive counseling is an established component of reproductive health services, and starting a method as part of post-abortion care is standard practice.⁷ International guidance emphasizes voluntary, informed, and non-stigmatising choice from a full range of methods after abortion.⁸ It is often assumed that women who begin a method after an abortion are more strongly motivated to avoid a further pregnancy, and services sometimes act on that assumption when deciding who receives closer follow-up. Recorded initiation context, however, is an administrative label; age, parity, and number of prior abortions are not measurements of preference or motivation, and whether the label identifies a group with different continuation behavior is an empirical question, not a given.

A recent prospective inception cohort in Shanghai followed 1156 women who began NET-EN/E2V after abortion and reported a mean use duration of 5.73 months and a one-year discontinuation rate of 98.72%.⁹ That study describes what happens from the first injection, and its central message is that attrition is early and near-universal. It leaves open a complementary question with direct service relevance: among the minority of women who do get past the first months, how long does an episode of use last, and what reasons do clinical records give when it ends?

We addressed that question in a cohort defined by at least six recorded injections. We selected six injections as the entry criterion for three reasons. First, six months is the point by which the Shanghai cohort had already lost approximately 93% of initiators, so the threshold separates the early-attrition phase from what follows.⁹ Second, receiving six recorded injections requires repeated engagement with a monthly, provider-administered service and therefore identifies women who continued beyond the earliest phase of use. Third, six completed visits indicate that the woman and the service had established a recurring delivery routine, which is what we mean by established use. The threshold is pragmatic, not biologically exact, and we use the term established user throughout to describe this record-defined group only.

Our objectives were to describe, among established NET-EN/E2V users: use status at the data cut-off; completed-use duration among women no longer active; recorded menstrual change and paired hemoglobin values; and the final reason for stopping documented in routine care. We also compared recorded post-abortion with other initiation contexts, to find out whether this routinely available label identifies women with different continuation, instead of assuming that it does, and we explored whether routinely recorded characteristics predicted completed use of at least 12 months.

Materials and Methods

Study Design and Setting

We conducted a retrospective cohort study in the Department of Family Planning of Wuxi Maternal and Child Health Hospital (Affiliated Hospital of Jiangnan University), a tertiary maternal and child health hospital in the city of Wuxi, Jiangsu Province, eastern China. The department provides family planning services to the city and its surrounding districts, including contraceptive counseling, post-abortion contraceptive care, and provider-administered injectable contraception. Eligible participants initiated NET-EN/E2V between May 2023 and November 2025, and routine injection and follow-up records were abstracted through May 2026. Reporting follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) recommendations.¹⁰

Ethics

The study was approved by the Institutional Review Board of Wuxi Maternal and Child Health Hospital (Approval No. YLSL2026-011) and was conducted in accordance with the Declaration of Helsinki. Because the study used retrospective, de-identified clinical records, the Board waived written informed consent. Telephone information came solely from routine clinical follow-up completed as part of usual care and documented before research abstraction; no participant was contacted or interviewed specifically for this study.

Participants

Eligible women were aged 18–45 years and had received at least six recorded NET-EN/E2V injections at the study center. The database included 134 eligible established users. The six-injection threshold, and the sense in which it stands in for established use, are explained in the Introduction; in brief, it separates early attrition from later use and identifies women who had returned repeatedly for a monthly provider-administered method. It is a pragmatic record-based marker of established use, not a biological threshold. Injections were not required to be strictly consecutive because routine records included temporary interruptions and restarts. Women with fewer than six recorded injections and those with critically incomplete baseline information were excluded.

We did not reconstruct the source population of all initiators, and women who stopped before six injections fell outside the analytic cohort. The denominator therefore cannot yield six- or 12-month continuation from initiation. No prospective sample-size calculation was performed; all eligible records in the database were analyzed.

Data Sources and Variable Definitions

Data were obtained from the hospital information system, injection records, outpatient notes, and routine follow-up logs. The source spreadsheet contained structured fields and free text. Individual injection-date fields included non-standard text, temporary interruptions, restarts, injections given elsewhere, and occasional entry irregularities. We checked summary fields against participant status and reason fields before analysis. Because exact intervals could not be reconstructed reliably for all participants, the main analyses used the curated summary status and duration fields in the database. No patient-identifying information was included in analytic outputs or the manuscript.

Variables included age at initiation, number of prior abortions, number of living children, initiation context, status at data cut-off, summary duration, self-reported menstrual blood-loss change, paired hemoglobin values when clinically available, and the documented reason for stopping. Initiation context was classified from the recorded reason as post-abortion initiation (PAI: post-abortion use with a menstrual or pain-related indication or contraception) or non-post-abortion initiation (NPAI: contraceptive counseling or method change). This label describes the clinical record and was not treated as a measure of motivation. We examined it because it is one of the few variables available to a family planning service at the moment of initiation: if post-abortion initiation identified women with systematically different continuation, services could use it to target closer follow-up, whereas if it does not, follow-up intensity has to be decided on other grounds.

Ascertainment of Reasons for Stopping

Reasons for stopping were not collected for research purposes. They were entered during usual care by the nursing and medical staff of the family planning clinic, through one of three routes: a woman attending for an injection or another visit said that she intended to stop, and the reason was entered in the outpatient note; a woman telephoned the clinic to cancel or to ask about stopping, and the call was logged in the follow-up record; or the clinic’s routine telephone call after a missed scheduled injection reached the woman and her reason was entered in the follow-up log. Entries such as relocation and forgetfulness therefore reached the record as the woman’s own account, given to clinic staff at the time and most often during that routine missed-appointment call; they were not inferred by the research team. Free-text entries were subsequently mapped to the predefined categories described under Outcome Definitions, and entries that did not clearly correspond to a category were left uncategorized, not forced into one.

We did not contact the women whose records held no documented reason. The study was approved as a retrospective analysis of existing de-identified records with a waiver of written informed consent; contacting participants for research purposes fell outside that approval and would have required a separate protocol and consent process. In addition, a reason recalled up to three years after stopping, elicited by a call from the treating institution, would be subject to recall and social-desirability bias and would not be comparable with the contemporaneous routine entries available for the other women. We therefore report documented reasons as documented reasons, and show the undocumented cases explicitly in every denominator.

Outcome Definitions

Women whose summary field indicated ongoing use were classified as active at data cut-off (n=36). A numeric summary duration was available for the 98 women no longer active and was analyzed only in that subgroup. Completed-use duration of at least 12 months served as an exploratory marker of longer use. A final stopping reason was available for 61 of the 98 non-active women. Four active women had a recorded reason related to an earlier interruption or restart; these historical entries were excluded from the final-reason analysis. When a record described more than one use episode, the most recent episode determined the final-reason category; earlier events were described separately but were not counted as final reasons. Reasons were grouped as practical barriers (clinic inconvenience, forgetfulness, or relocation), menstrual changes, other side effects, or contraceptive failure. Elevated blood-pressure readings were tracked separately within the other-side-effects group because blood-pressure assessment is clinically relevant for an estrogen-containing method.

Statistical Analysis

Continuous variables are reported as mean±standard deviation (SD) and categorical variables as n (%). Baseline characteristics were compared between PAI and NPAI participants using independent-samples t-tests, and active status using the Pearson chi-square test. Numeric-duration comparisons were restricted to the 98 non-active women. Duration by menstrual-change category was assessed by one-way analysis of variance, and paired hemoglobin values by a paired t-test. Missing values were not imputed, and denominators are reported for each analysis.

No prospective sample-size calculation was performed, because the analysis used all eligible records in the database, not a recruited sample. We therefore report 95% confidence intervals (CIs) for the principal proportions, calculated by the Wilson method, so that the precision actually achieved is explicit. With 98 women contributing to duration analyses, a proportion near 40% is estimated with a half-width of roughly 10% points, and with 61 documented reasons a proportion near 50% with a half-width of roughly 12 points. The study can identify the most frequently recorded category, but it is not powered to detect modest differences between subgroups; the subgroup percentages by duration are therefore presented as descriptive only.

Exact injection dates could not be reconstructed reliably for all participants, so time-to-event methods such as Kaplan–Meier estimation and Cox regression were not appropriate and were not attempted. All duration results are descriptive summaries of completed episodes among women who were no longer active, and none of them can be read as a survival estimate.

An exploratory multivariable logistic regression was fitted among the 98 non-active women, with completed-use duration ≥12 months as the outcome and age (per year), number of prior abortions (per abortion), and PAI status specified a priori as the three candidate predictors. Adjusted odds ratios (aORs), 95% CIs, and P values are reported, with model performance summarized by the omnibus likelihood-ratio test, Nagelkerke R², and the area under the receiver operating characteristic curve (AUC). With 40 outcome events the model was deliberately capped at three predictors and used descriptively, not for individual prediction or causal inference. Analyses were run in SPSS 26.0 and independently reproduced from the original spreadsheet in Python. Two-sided P<0.05 was considered significant; because the analyses were exploratory and involved several comparisons, we emphasize effect estimates and confidence intervals over isolated P values.

Results

Participant Characteristics

The 134 participants had a mean age of 31.9±5.1 years (range 19–43). Eighteen (13.4%) were aged ≤25 years, 34 (25.4%) were 26–30 years, 47 (35.1%) were 31–35 years, and 35 (26.1%) were ≥36 years. The mean number of prior abortions was 1.91±1.70 and of living children 1.19±0.86. PAI was recorded for 63 women (47.0%) and NPAI for 71 (53.0%). At data cut-off, 36 (26.9%; 95% CI 20.1–34.9) were active and 98 (73.1%) were not (Table 1).

Table 1.

Characteristics and Status of Established NET-EN/E2V Users (n=134)

Variable	Value
Age, years, mean±SD	31.87±5.14
Age distribution, n (%)
≤25 years	18 (13.4)
26–30 years	34 (25.4)
31–35 years	47 (35.1)
≥36 years	35 (26.1)
Prior abortions, mean±SD	1.91±1.70
Living children, mean±SD	1.19±0.86
Initiation context, n (%)
Post-abortion initiation	63 (47.0)
Menstrual/pain-related indication	45 (33.6)
Primarily for contraception	18 (13.4)
Non-post-abortion initiation	71 (53.0)
Status at data cut-off, n (%)
Active	36 (26.9)
Not active	98 (73.1)

Abbreviations: NET-EN/E2V, norethisterone enanthate 50 mg/estradiol valerate 5 mg; SD, standard deviation.

Initiation Context and Recorded Duration

PAI participants were older than NPAI participants (32.9±4.7 vs 30.9±5.4 years; P=0.026) and had more prior abortions (2.27±1.86 vs 1.59±1.48; P=0.020); living children did not differ. Active use at data cut-off was recorded for 19/63 PAI (30.2%) and 17/71 NPAI participants (23.9%; P=0.418). Among women no longer active, numeric duration was available for 44 PAI and 54 NPAI participants, with mean durations of 11.43±5.97 and 11.87±5.76 months (mean difference −0.44 months, 95% CI −2.80 to 1.92; P=0.713) (Table 2).

Table 2.

Comparison by Recorded Initiation Context

Variable	PAI (n=63)	NPAI (n=71)	Statistic	P
Age, years	32.92±4.67	30.94±5.38	t=2.257	0.026
Prior abortions	2.27±1.86	1.59±1.48	t=2.348	0.020
Living children	1.25±0.97	1.13±0.75	t=0.854	0.395
Active at cut-off, n (%)	19 (30.2)	17 (23.9)	χ2=0.656	0.418
Duration among non-active, months	11.43±5.97 (n=44)	11.87±5.76 (n=54)	t=−0.369	0.713

Notes: Values are mean±SD or n (%). Numeric duration was analyzed only among women not active at data cut-off (PAI n=44; NPAI n=54); the mean difference (PAI minus NPAI) was −0.44 months (95% CI −2.80 to 1.92).

Abbreviations: CI, confidence interval; NPAI, non-post-abortion initiation; PAI, post-abortion initiation; SD, standard deviation.

Completed-Use Duration and Status at Data Cut-Off

Among the 98 non-active women, mean completed-use duration was 11.7±5.8 months and the median was 10 months (interquartile range 7–13.75). Forty (40.8%; 95% CI 31.6–50.7) had reached at least 12 months, 14 (14.3%) at least 18 months, and six (6.1%) at least 24 months, the longest being 33 months. The full distribution appears in Figure 1A and Table 3. The 36 active women were not assigned numeric duration thresholds, because their summary field recorded ongoing status instead of a standardized value, so their still-accruing durations are not captured in these figures.

Figure 1.

Completed-use duration and documented final reasons for stopping among women no longer active at data cut-off. (A) Distribution of recorded completed-use duration among 98 non-active women; the dashed line marks the ≥12-month threshold (40/98, 40.8%). (B) Final reasons (column percentages) among the 61 women with a documented reason, grouped by completed-use duration. Practical barriers accounted for 43.5% of documented reasons at 6–11 months and 66.7% at ≥12 months.

Table 3.

Recorded Completed-Use Duration Among Women Not Active at Data Cut-Off (n=98)

Measure	Value
Duration, months, mean±SD	11.67±5.83
Duration, months, median (IQR)	10 (7–13.75)
6 months, n (%)	15 (15.3)
7–9 months, n (%)	27 (27.6)
10–11 months, n (%)	16 (16.3)
12–17 months, n (%)	26 (26.5)
18–23 months, n (%)	8 (8.2)
≥24 months, n (%)	6 (6.1)
Duration ≥12 months, n (%) [95% CI]	40 (40.8) [31.6-50.7]
Duration ≥18 months, n (%)	14 (14.3)

Notes: Numeric duration was available only for women not active at data cut-off; percentages use n=98 as the denominator. The 95% CI is calculated by the Wilson method.

Abbreviations: CI, confidence interval; IQR, interquartile range; SD, standard deviation.

Menstrual Changes and Hemoglobin

Across all 134 participants, menstrual blood loss was recorded as decreased in 102 (76.1%), unchanged in 26 (19.4%), and increased in six (4.5%). Among the 98 non-active women with numeric duration, the category sizes were 69, 24, and five. Mean duration was 12.33±5.92 months for decreased blood loss, 10.42±5.85 months for unchanged, and 8.60±1.52 months for increased (P=0.185), and at least 12 months of use occurred in 33/69 (47.8%), 7/24 (29.2%), and 0/5, respectively. Paired hemoglobin values were available for 17 women, with mean concentration moving from 132.0±8.4 to 134.2±8.2 g/L (mean paired difference 2.24±7.88 g/L; P=0.259) (Table 4).

Table 4.

Menstrual Blood-Loss Change, Duration Among Non-Active Women, and Hemoglobin

Measure	Result	P
Menstrual blood-loss change, all (n=134)	Decreased 102 (76.1); unchanged 26 (19.4); increased 6 (4.5)	—
Duration by menstrual category (non-active), months	Decreased 12.33±5.92 (n=69); unchanged 10.42±5.85 (n=24); increased 8.60±1.52 (n=5)	0.185
Duration ≥12 months by category, n/N (%)	Decreased 33/69 (47.8); unchanged 7/24 (29.2); increased 0/5 (0.0)	—
Paired hemoglobin subsample	n=17	—
Pre-use Hb, g/L, mean±SD	132.0±8.4	—
Post-use Hb, g/L, mean±SD	134.2±8.2	—
Mean paired change, g/L, mean±SD	2.24±7.88	0.259

Notes: Duration analyses included non-active women with numeric duration (decreased n=69; unchanged n=24; increased n=5). The P values are from one-way analysis of variance (duration) and a paired t-test (hemoglobin).

Abbreviation: Hb, hemoglobin; SD, standard deviation.

Documented Reasons for Stopping

A documented final reason was available for 61 of the 98 non-active women (62.2%; 95% CI 52.4–71.2); the remaining 37 women had no reason recorded and are retained in all duration denominators but excluded from all reason denominators. Practical barriers accounted for 30/61 documented reasons (49.2%; 95% CI 37.1–61.4): clinic inconvenience (n=20), forgetfulness (n=8), and relocation (n=2). Menstrual changes accounted for 18/61 (29.5%; 95% CI 19.6–41.9), other side effects for 10/61 (16.4%), and contraceptive failure for 3/61 (4.9%). Five of the ten other-side-effect records noted elevated or borderline-elevated blood-pressure readings (systolic 125–150 mmHg and diastolic 85–96 mmHg); the remaining records noted breast tenderness (n=3), nausea (n=1), or acne (n=1). One of the three final contraceptive failures followed a documented off-schedule injection at an outside facility. A separate participant had an earlier failure after mistimed injection, restarted the method, and later stopped because of menstrual change; that earlier event was not counted as her final reason. Four active women had historical interruption or restart reasons and were excluded, as specified in Methods (Table 5).

Table 5.

Documented Final Reasons for Stopping Among Non-Active Women (n=61)

Reason	6–11 months (n=46)	≥12 months (n=15)	Total (n=61)
Practical barriers	20 (43.5)	10 (66.7)	30 (49.2)
Clinic inconvenience	14	6	20 (32.8)
Forgetfulness	5	3	8 (13.1)
Relocation	1	1	2 (3.3)
Menstrual changes	16 (34.8)	2 (13.3)	18 (29.5)
Other side effects	7 (15.2)	3 (20.0)	10 (16.4)
Elevated or borderline-elevated blood-pressure readings	3	2	5 (8.2)
Other (breast tenderness, nausea, acne)	4	1	5 (8.2)
Contraceptive failure	3 (6.5)	0 (0.0)	3 (4.9)

Notes: Percentages are column percentages among the 61 non-active women with a documented final reason; 37 additional non-active women had no documented reason and were excluded. Reasons came from usual-care records (see Ascertainment of Reasons for Stopping). One final failure followed an off-schedule injection at another facility. An earlier failure after mistimed injection was not counted because the participant restarted and later stopped for menstrual change. Four active women with historical interruption or restart reasons were excluded. No prespecified inferential comparison was made between duration groups.

All of these entries originated in routine clinical documentation as set out under Ascertainment of Reasons for Stopping; the practical-barrier entries, including relocation and forgetfulness, were statements made by the women themselves to clinic staff, most often during the clinic’s routine telephone call after a missed injection.

The distribution of final reasons differed descriptively by completed-use duration (Figure 1B). Among 46 women who stopped at 6–11 months and had a documented reason, practical barriers accounted for 20 (43.5%) and menstrual changes for 16 (34.8%). Among 15 women who stopped at ≥12 months, practical barriers accounted for 10 (66.7%) and menstrual changes for two (13.3%). These small, post hoc subgroups were not compared inferentially; given 15 women in the longer-duration group, the percentages are imprecise and should be treated as descriptive.

Exploratory Predictors of Completed Use ≥12 Months

Among the 98 non-active women, 58 stopped at 6–11 months and 40 reached at least 12 months. None of the three prespecified predictors was statistically significant: age (aOR 0.95 per year, 95% CI 0.87–1.04; P=0.265), prior abortions (aOR 0.99 per abortion, 95% CI 0.77–1.28; P=0.930), and PAI status (aOR 0.81, 95% CI 0.35–1.90; P=0.631). Overall discrimination was poor (omnibus χ²=2.264, df=3, P=0.519; Nagelkerke R²=0.031; AUC=0.580) (Table 6). Given the wide confidence intervals, these results argue against these routinely recorded variables being strong markers of longer use; they do not exclude a weaker association.

Table 6.

Exploratory Logistic Regression for Completed-Use Duration ≥12 months Among Non-Active Women (n=98)

Variable	6–11 mo (n=58)	≥12 mo (n=40)	Unadj. P	Adjusted OR (95% CI); P
Age, years, mean±SD	32.97±4.99	31.45±5.51	0.160	0.95 (0.87–1.04); 0.265
Prior abortions, mean±SD	2.17±1.96	1.90±1.68	0.475	0.99 (0.77–1.28); 0.930
Post-abortion initiation, n (%)	28 (48.3)	16 (40.0)	0.418	0.81 (0.35–1.90); 0.631

Notes: Age and prior abortions were continuous; post-abortion initiation was binary. Model: omnibus χ²=2.264 (df=3), P=0.519; Nagelkerke R²=0.031; AUC=0.580.

Abbreviations: AUC, area under the ROC curve; CI, confidence interval; OR, odds ratio; SD, standard deviation.

Discussion

Principal Findings

This retrospective cohort describes 134 established users of monthly NET-EN/E2V. At data cut-off, 36 remained active; among the 98 who were no longer active, 40.8% had completed at least one year of use and six had completed at least two years. Among the 61 women whose records documented a final reason, practical barriers were the most common category, followed by menstrual changes, and they accounted for a larger share of documented reasons among women with longer completed use. Recorded post-abortion initiation was associated with neither active status nor completed duration. The exploratory model discriminated poorly and identified no statistically significant predictor of use ≥12 months. These findings apply to a selected group that had already received six injections and should not be read as continuation rates from initiation.

Comparison with Previous Studies

Our results can be set against three kinds of earlier evidence. Under trial conditions, NET-EN/E2V performed well: in a multicenter Phase III program covering 15,787 woman-months in eight Latin American centers, continuation was 64.7% at 12 months and 31.0% at 24 months, and cumulative discontinuation was 6.1% for bleeding problems and 7.2% for other adverse events at 36 months.⁵ A subsequent multi-country program evaluation of the same formulation reported an overall one-year discontinuation rate of 17.9%, of which 5.1% was attributable to bleeding problems.⁴ A Cochrane review of CICs likewise identified bleeding disturbance as the principal drug-related reason for stopping.⁶ Second, outside trial conditions continuation is markedly lower: among Chilean adolescents followed for 12 months in ordinary clinical practice, 41.9% of NET-EN/E2V users were still using the method at one year.¹¹ Third, an early review of the Phase III literature on monthly injectables noted that alongside abnormal bleeding, which was the main drug-related complaint, missed appointments recurred as a reason for stopping, and read this as a signal that the monthly schedule itself imposes a burden on users and services.¹²

Our data are consistent with that observation. Among women who had passed the six-injection threshold and whose records documented a reason, practical access barriers were the most frequently recorded category overall (49.2%). They accounted for 43.5% of documented reasons at 6–11 months and 66.7% at ≥12 months. Only 15 women were in the latter group, so the subgroup difference is descriptive and imprecise. These proportions are not directly comparable with the trial figures above, which are cumulative life-table discontinuation rates from initiation rather than shares of documented reasons within a selected, later-stage cohort. The rank order can nevertheless be compared descriptively: practical barriers were recorded more often than bleeding in our cohort.

A Complement to the Shanghai Inception Cohort

The Shanghai study and the present study address different parts of contraceptive use. The Shanghai cohort followed post-abortion initiators from the first injection and reported 93% cumulative discontinuation by six months and 98.72% by one year.⁹ Our cohort was restricted to women who had already received at least six injections and describes completed duration and documented reasons for later stopping. The studies therefore describe different stages of use rather than points on a directly comparable curve. Taken together, their findings suggest, but do not establish, that reasons for stopping may differ by stage of use. An inception cohort that records reasons throughout follow-up would be needed to test that possibility. Direct comparison of percentages is inappropriate because eligibility criteria, follow-up, and outcome definitions differ.

Interpretation of Initiation Context

PAI participants were older and reported more prior abortions than NPAI participants, but such differences do not establish greater motivation, which, along with preference, counseling quality, and decision processes, was not measured. The comparison is useful because it was null. Recorded initiation context is one of the few variables a family planning service holds for every woman at the moment she starts the method, and it would be operationally convenient if it flagged those needing closer follow-up. In this cohort it did not: active status and completed duration were similar across contexts despite the baseline differences in age and prior abortions. This study therefore did not identify post-abortion initiation as a useful standalone marker for follow-up intensity or as a proxy for motivation. This should be read as an absence of evidence for a difference in a selected cohort of moderate size, not as demonstrated equivalence.

Menstrual Experience and Clinical Monitoring

Most women had a routine-care record of decreased menstrual blood loss, but this was a simple self-reported category rather than a validated bleeding diary or pictorial assessment. Category-specific duration analyses were confined to non-active women, and only five women with increased flow had a numeric duration. The findings are therefore descriptive and cannot establish that menstrual change influenced continuation. The paired hemoglobin analysis, based on 17 clinically selected women, showed no statistically significant change and demonstrates neither hematologic benefit nor harm. Prospective studies should prespecify measurement points, use standardized bleeding outcomes, and collect paired laboratory data systematically. Separately, five discontinuation records noted elevated or borderline-elevated blood-pressure readings. These observations were not based on a standardized monitoring protocol and do not establish causation, but they reinforce the need for guideline-concordant blood-pressure assessment during use of an estrogen-containing method.

Service Implications

Clinic inconvenience, forgetfulness, and relocation accounted for almost half of documented final reasons, with a higher descriptive proportion among women who had used the method for at least 12 months. For a monthly provider-administered method, appointment reminders, flexible scheduling, coordination with other visits, and easier transfer between clinics could be evaluated. A text-message pilot improved attendance for injectable contraception in the short term, although the effect was not sustained at later visits.¹³ Any intervention should be tested specifically for NET-EN/E2V and within the local service context, and the case for testing them rests on the documented-reason subgroup rather than on the cohort as a whole.

Self-administration has improved continuation of subcutaneous depot medroxyprogesterone acetate in prospective evidence, and WHO guidance supports self-administration of DMPA-SC as an additional delivery option.^14,15 That evidence does not transfer to intramuscular NET-EN/E2V: any move toward self-injection would require product-specific evidence, training, safety assessment, supply arrangements, and regulatory approval. The immediate, actionable implication of our data is narrower, namely reducing avoidable access and scheduling barriers within existing, approved delivery pathways.

Strengths and Limitations

The study uses participant-level routine-care data, distinguishes active from non-active status, restricts duration analyses to women with a numeric summary value, separates historical interruption reasons from final reasons, and reports the provenance of every stopping reason. Its focus on established users complements evidence from cohorts that begin at method initiation.

Several limitations temper these findings. The six-injection entry criterion created survivor selection and excluded early discontinuers, so the study cannot estimate continuation from initiation, and every proportion reported here is conditional on having reached that threshold. Numeric duration was unavailable in standardized form for the 36 active women, restricting duration thresholds and regression to the 98 non-active women and truncating the upper end of the duration distribution.

A final reason was missing for 37 of the 98 non-active women (37.8%), and this is the most consequential limitation for our main claim. Missingness is unlikely to be random, but its direction is not obvious. A woman who simply stops attending and cannot be reached by the routine follow-up call is, by the mechanism of ascertainment, plausibly more likely to have stopped for a logistical reason, which would mean our estimate understates practical barriers; equally, she may have switched to another method elsewhere without difficulty, which would mean the opposite. Because the direction cannot be resolved with these data, the reported percentages should be read as the distribution of documented reasons among the 61 women who have one, not as the distribution of reasons among all women who stopped. The conclusion is therefore framed throughout in terms of documented reasons.

Reasons also came from routine documentation, not from a validated discontinuation interview, and free-text records included interruptions, restarts, and entry irregularities, creating a risk of misclassification and preventing reliable reconstruction of exact injection intervals or any survival analysis. No prospective power calculation was performed, and the confidence intervals reported in the Results show the resulting imprecision, which is substantial for the duration subgroups. Menstrual change was self-reported, blood-pressure measurements were not collected under a standardized research protocol, hemoglobin data were available for only 17 women, and socioeconomic, counseling, relationship, and preference variables were unavailable. The single-center sample and small subgroups also limit precision and generalizability. Future multicenter inception cohorts should enroll all initiators, define allowable injection windows, distinguish temporary interruption from permanent discontinuation, and collect reasons at prespecified follow-up points.

Conclusions

Among women who had already received at least six monthly NET-EN/E2V injections and whose records documented a reason for stopping, practical access barriers were the most frequently recorded reason and accounted for a larger descriptive share after longer completed use. A reason was documented for 61 of the 98 women who were no longer active, so this describes documented reasons rather than all discontinuations. Recorded post-abortion initiation was associated with neither active status nor completed-use duration and was not identified as a useful standalone marker for follow-up intensity in this cohort. Because early discontinuers were excluded by design and active users lacked standardized numeric duration, these findings do not estimate continuation from initiation. Prospective inception-cohort studies that collect reasons at prespecified time points are needed to test whether service interventions that reduce access and scheduling burdens improve continuation among established users.

Funding Statement

This work was supported by the Key Discipline of Maternal and Child Health in Jiangsu Province (Third Cycle) (SFY3-JS2021) and the General Research Project on Maternal and Child Health of Wuxi Municipal Health Commission (FYKY202303). The funders had no role in study design; data collection, analysis, or interpretation; manuscript preparation; or the decision to submit.

Use of AI-Assisted Technology statement

During manuscript preparation, the authors used Claude Opus 4.8 (Anthropic) and ChatGPT with GPT-5.6-sol (OpenAI) solely for English-language editing and formatting of author-prepared text. These tools were not used to generate study data, perform statistical analyses, or create references. The authors reviewed and verified all AI-assisted changes and take full responsibility for the accuracy, originality, and integrity of the final manuscript, including its references. The authors confirm that the terms of use of the specific AI tool were checked and therefore confirm suitability for publication.

Abbreviations

aOR, adjusted odds ratio; AUC, area under the receiver operating characteristic curve; CI, confidence interval; CIC, combined injectable contraceptive; DMPA-SC, subcutaneous depot medroxyprogesterone acetate; NET-EN/E2V, norethisterone enanthate 50 mg/estradiol valerate 5 mg; NPAI, non-post-abortion initiation; OR, odds ratio; PAI, post-abortion initiation; SD, standard deviation; STROBE, Strengthening the Reporting of Observational Studies in Epidemiology; WHO, World Health Organization.

Data Sharing Statement

The de-identified dataset is not publicly available because it contains sensitive clinical information subject to institutional data-governance restrictions. It may be available from the corresponding author on reasonable request and subject to approval by the Institutional Review Board of Wuxi Maternal and Child Health Hospital.

Ethics Approval and Informed Consent

The study was approved by the Institutional Review Board of Wuxi Maternal and Child Health Hospital (Approval No. YLSL2026-011) and conducted in accordance with the Declaration of Helsinki. Because it used retrospective, de-identified clinical records and routine follow-up documentation without research-specific participant contact, the Board waived written informed consent. Telephone follow-up data came exclusively from routine clinical records generated before the study; no additional interviews or participant contact were conducted for research.

Consent for Publication

Not applicable. The manuscript contains no identifiable personal information, images, videos, or recordings.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare that they have no financial or non-financial competing interests relevant to this work.