Ethics statement

Ethical approval for this study was obtained from the Kampala International University Research Ethics Committee (KIU-REC) under approval number KIU-REC-2025–1732. Administrative permission to conduct the study was granted by Jinja Regional Referral Hospital, Lira Regional Referral Hospital, and Kayunga Regional Referral Hospital. The study was also registered with the Uganda National Council for Science and Technology (UNCST).

Written informed consent was obtained from all participants before enrolment after providing detailed information about the study in a language they understood. Participation was entirely voluntary, and all participants were aged 18 years or older and provided consent personally. Confidentiality was maintained through the use of unique study identification numbers and secure storage of questionnaires, laboratory records, and other study-related documents. The study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. Participants with positive cultures were counselled and referred for treatment with cefixime 400 mg orally as a single dose or ceftriaxone 1 g intravenously as a single dose, plus doxycycline 100 mg twice daily for seven days and metronidazole 2 g orally as a single dose. Erythromycin replaced doxycycline during pregnancy. Partner evaluation and treatment were encouraged to prevent reinfection

Study design and setting

We conducted a multicenter, hospital-based cross-sectional study at Jinja, Lira, and Kayunga Regional Referral Hospitals between 10^th September 2025 and 10^th January 2026. These public tertiary hospitals serve different geographical regions of Uganda. Jinja Regional Referral Hospital is a 500-bed facility serving approximately 4.6 million people in Jinja City and 11 surrounding districts. Lira Regional Referral Hospital has 401 beds and serves approximately 2.5 million people across nine districts and Lira City. Kayunga Regional Referral Hospital serves more than 7 million people across six districts in central Uganda.

All three hospitals receive referrals from district hospitals and lower-level health facilities and provide specialist gynecology and fertility services through outpatient clinics. Their microbiology laboratories can perform bacterial culture, organism identification, and routine antimicrobial susceptibility testing. However, long travel distances, transport costs, hospital congestion, and out-of-pocket costs for investigations may limit access for some women. The hospitals were purposively selected because they had established fertility services, functional microbiology laboratories, and served populations from different regions of Uganda, thereby broadening the geographical relevance of the study findings. Jinja, Kayunga and Lira Regional Referral Hospitals are located in eastern, central and northern Uganda, respectively. Based on their geographical coordinates, the hospitals are separated by approximately 46–205 km in straight-line distance and serve geographically and socially diverse urban and rural referral populations.

Study period and participants

Data were collected from September 10, 2025, to January 10, 2026. Eligible participants were women presenting for the evaluation of primary or secondary infertility after at least 12 months of regular unprotected sexual intercourse. Women were included if they met the infertility definition and provided written informed consent. The exclusion criteria were refusal to participate, menstruation at the time of sampling, prior hysterectomy, inability to undergo speculum examination, and incomplete study data. Women reporting oral, injectable or intravaginal/topical antibiotic use during the preceding two weeks were also excluded because recent antimicrobial exposure could suppress gonococcal growth and reduce culture sensitivity.

Sample size and sampling

The sample size was estimated using the single population proportion formula described by Kish and Leslie (1965): n = Z² p (1 − p)/d², where n is the required sample size, Z is the standard normal deviate at a 95% confidence level (1.96), p is the estimated prevalence, and d is the margin of error, set at 0.05. Using an estimated Neisseria gonorrhoeae prevalence of 8.4% reported in Tanzania [35], the initial sample size was calculated as follows:

After allowing 10% for nonresponse, the minimum target was approximately 130 participants. This calculation did not incorporate a design effect because a reliable estimate of within-hospital correlation was unavailable during planning. Recruitment continued throughout the prespecified study period, resulting in a final sample of 402 women, with 134 recruited from each hospital.

On each recruitment day, potentially eligible women were identified sequentially from the fertility or gynecology clinic attendance register. Following selection of a random starting position from the first three eligible attendees, every third eligible woman was approached by a trained research assistant in a private area after routine clinic registration. Eligibility was confirmed, and the study was explained in the participant’s preferred language. Written informed consent was obtained before administration of the questionnaire or collection of an endocervical specimen. Women who declined participation continued to receive routine clinical care without disadvantage.

Data collection procedures

After providing written informed consent, participants were interviewed by trained bilingual research assistants using a structured, interviewer-administered questionnaire. The questionnaire collected sociodemographic, sexual and behavioural, reproductive, and relevant clinical information. Because no single standardized instrument addressed all the study objectives, the questionnaire was developed from the study objectives and informed by previous studies on gonococcal infection, infertility, and related reproductive-health factors. It was reviewed for relevance and clarity and pilot-tested before the main study. Findings from the pilot test were used to refine the wording, response options, and sequence of questions.

The questionnaire was administered in English or the participant’s preferred local language. Interviews were conducted in Lusoga at Jinja, Luganda at Kayunga, and Lango at Lira by research assistants fluent in the respective languages. All research assistants received standardized training in participant recruitment, informed consent, interviewing, questionnaire completion, confidentiality, and accurate recording of responses. The same questionnaire and interview procedures were used across the three hospitals to reduce interviewer-related variation.

Consistent eligibility criteria and systematic recruitment were used to reduce selection bias. Because sensitive behavioural and reproductive information was self-reported, recall and social-desirability bias remained possible. Interviews were therefore conducted privately using neutral, non-judgmental language. Standardized procedures for specimen collection, transportation, culture, identification, and antimicrobial susceptibility testing were applied across the study sites to reduce laboratory variation and potential misclassification. Before the main study, the questionnaire was pilot-tested among 40 women, representing approximately 10% of the planned sample

Outcome and covariates

Infertility was defined according to the World Health Organization as failure to achieve pregnancy after at least 12 months of regular unprotected sexual intercourse. Time since infertility diagnosis was considered a separate variable and was defined as the participant-reported interval between clinical confirmation of infertility and enrolment in the study. It was categorized as <1 year, 1–3 years, and >3 years to distinguish recently diagnosed, intermediate, and longer-standing infertility. Therefore, the < 1-year category referred to time since clinical diagnosis and not to the duration of attempts to conceive.

The primary outcome was gonococcal cervicitis, defined as the isolation of Neisseria gonorrhoeae from an endocervical swab using culture and standard microbiological identification procedures. Prespecified explanatory variables included sociodemographic characteristics (age, marital status, educational level, occupation, and monthly household income); sexual and behavioural characteristics (alcohol use, age at first sexual intercourse, lifetime number of sexual partners, a new sexual partner during the preceding three months, and vaginal douching during the preceding three months); obstetric and gynaecological characteristics (abnormal vaginal discharge during the preceding three months, previous ectopic pregnancy, oral contraceptive use, and time since infertility diagnosis); and clinical characteristics (HIV status as per patient records, history of anaemia, and body mass index [BMI]).

These variables were selected before analysis based on published epidemiological evidence, biological plausibility, and their relevance to gonococcal exposure or reproductive morbidity. Sexual-partnership variables were included as measures of potential exposure, while HIV status was considered because HIV and gonococcal infection may occur within overlapping sexual networks. Vaginal douching was included because of its potential effect on the cervicovaginal environment. Previous ectopic pregnancy and time since infertility diagnosis were included to reflect the participant’s reproductive history, while abnormal vaginal discharge was included as a possible clinical presentation of cervical infection, recognizing that gonococcal infection in women may also be asymptomatic.

Age was categorized as <20, 20–29, 30–39, and ≥40 years to provide clinically meaningful reproductive-age groups while retaining sufficient observations for analysis. Formal employment referred to salaried work within a government institution, registered private organization, or another formally constituted employer. Informal employment included small-scale trading, subsistence farming, casual labour, domestic work, and self-employment without a formal contract. Participants without a current income-generating occupation were categorized as unemployed. Monthly household income was grouped using prespecified thresholds of 55USD and 110USD.

Weight was measured using a calibrated weighing scale, and height was measured using a stadiometer. BMI was calculated as weight in kilograms divided by height in meters squared and categorized according to WHO criteria as underweight (<18.5 kg/m²), normal weight (18.5–24.9 kg/m²), overweight (25.0–29.9 kg/m²), or obesity (≥30.0 kg/m²).

Vaginal douching was defined as the intentional introduction of water, another liquid, or a cleansing product into the vagina for cleansing or perceived hygiene at least once during the three months preceding enrolment. It was recorded as a binary variable; information on frequency, products used, volume, and method of application was not collected. Alcohol use was defined as the consumption of any alcoholic beverage during the preceding three months and was also recorded as a binary variable. The questionnaire did not distinguish occasional, weekly, daily, or heavy alcohol consumption.

Specimen collection and laboratory methods

Endocervical swabs were collected by trained clinicians using sterile specula and swabs under aseptic conditions. Each specimen was labelled with a unique study identification number and promptly transported to the onsite microbiology laboratory, where it was inoculated onto selective Modified Thayer–Martin agar. Culture plates were incubated at 35°C–37°C in approximately 5% carbon dioxide for 24–48 hours. The available study records did not document the transport medium, transport temperature, exact collection-to-inoculation interval, specimen-rejection threshold or use of a cold chain; therefore, these details could not be reconstructed retrospectively. Isolates were identified as Neisseria gonorrhoeae based on characteristic colony morphology, Gram-negative diplococci on Gram staining and a positive oxidase reaction.

Ceftriaxone was included because extended-spectrum cephalosporins remain central to effective gonorrhoea treatment. Azithromycin was included because of its historical role in combination therapy and the importance of monitoring emerging macrolide non-susceptibility. Ciprofloxacin, penicillin, and tetracycline were included to characterize resistance to previously use antimicrobial classes and to allow comparison with earlier Ugandan, regional, and global surveillance findings. Antibiotic selection was also guided by disc availability and the testing capacity of the participating laboratories.

Antimicrobial susceptibility testing was performed using the Kirby–Bauer disk-diffusion method on GC agar base supplemented with 1% defined growth supplement. A direct colony suspension equivalent to a 0.5 McFarland turbidity standard was uniformly inoculated onto the agar surface. Commercial antimicrobial discs containing ceftriaxone 30 µg, azithromycin 15 µg, ciprofloxacin 5 µg, tetracycline 30 µg, and penicillin 10 units were applied. Plates were incubated at 35°C–37°C in approximately 5% carbon dioxide for 20–24 hours, after which inhibition-zone diameters were measured in millimeters. GC agar base supplemented with 1% defined growth supplement is the standardized medium for gonococcal susceptibility testing. WHO laboratory guidance

Zone diameters were interpreted according to CLSI M100, 35th edition (2025). The susceptible, intermediate, and resistant thresholds were ≥47, 27–46, and ≤26 mm for penicillin; ≥ 41, 28–40, and ≤27 mm for ciprofloxacin; and ≥38, 31–37, and ≤30 mm for tetracycline, respectively. CLSI provides susceptible-only disk-diffusion thresholds of ≥35 mm for ceftriaxone and ≥30 mm for azithromycin. Isolates with zone diameters below these thresholds were therefore reported as below the susceptible-only threshold and were not classified as intermediate or resistant because MIC confirmation was not performed.

Neisseria gonorrhoeae ATCC 49226 was used for internal quality control. The applicable CLSI quality-control zone ranges were 39–51 mm for ceftriaxone, 30–38 mm for azithromycin, 48–58 mm for ciprofloxacin, 30–42 mm for tetracycline, and 26–34 mm for penicillin. Individual quality-control zone measurements and documentation of repeated testing runs were not available in the study dataset; therefore, the occurrence or absence of quality-control discordance could not be determined retrospectively. CLSI M100

Participants with positive gonococcal cultures were counselled and treated according to the Ugandan national STI treatment guidelines in force during the study period. Partner notification and treatment were encouraged to reduce the risk of reinfection. Because follow-up was outside the cross-sectional study procedures, routine microbiological test-of-cure was not undertaken. Culture was selected because viable isolates were required for phenotypic antimicrobial susceptibility testing. Nevertheless, culture is generally less sensitive than nucleic acid amplification testing, particularly when bacterial load is low or specimen transportation and processing conditions are suboptimal.

Quality control

Quality-control measures were applied throughout participant recruitment, questionnaire administration, specimen handling, laboratory testing, and data management. Before data collection, research assistants and laboratory personnel from the three study hospitals received standardized training on the study procedures. The same study definitions, questionnaire, specimen-collection procedures, and harmonized laboratory protocols were used across all three sites. Endocervical specimens were labelled with unique study identification numbers and promptly transported to the respective hospital laboratories under standardized conditions. Neisseria gonorrhoeae ATCC 49226 was used as the internal quality-control strain for antimicrobial susceptibility testing. Quality-control zone diameters were assessed against the CLSI acceptable ranges of 39–51 mm for ceftriaxone, 30–38 mm for azithromycin, 48–58 mm for ciprofloxacin, 30–42 mm for tetracycline, and 26–34 mm for penicillin. However, individual quality-control zone measurements and documentation of repeated testing runs were unavailable for retrospective review. Therefore, it was not possible to determine whether any out-of-range quality-control result or repeat testing occurred. Completed questionnaires were reviewed daily by the principal investigator for completeness and internal consistency. Data-entry records were cross-checked against the completed questionnaires, and regular supervision was conducted throughout the study to assess adherence to the approved protocol.

Data management and statistical analysis

Data were entered, cleaned, and analyzed using Stata version 17. Descriptive statistics were used to summarize participant characteristics. Prevalence of gonococcal cervicitis was calculated as the proportion of culture-confirmed cases among all tested participants. Participants with incomplete study data were excluded before analysis; accordingly, the final analytic dataset comprised 402 women with complete questionnaire and laboratory data. Bivariable logistic regression was used to examine the associations between explanatory variables and GC. Variables with p < .20 were entered into a multivariable logistic regression model. Adjusted odds ratios with 95% confidence intervals were reported. Multicollinearity was assessed using variance inflation factors based on the same indicator-variable coding used in the multivariable model. A VIF greater than 5 was considered potentially problematic. The maximum individual VIF was 3.20, the minimum tolerance was 0.313, and the maximum adjusted generalized VIF was 1.17. Pairwise associations among categorical predictors were weak, with a maximum Cramér’s V of 0.155. Therefore, no variable was removed because of multicollinearity. No sensitivity analyses were performed. Two prespecified multiplicative interactions were evaluated: HIV status × new sexual partner in the preceding three months and vaginal douching × abnormal vaginal discharge in the preceding three months. Each interaction was assessed separately by comparing nested main-effects and interaction models using a one-degree-of-freedom likelihood-ratio test. Both constituent main effects were retained during each interaction test. Because only 34 culture-positive outcomes were observed, the interaction analyses were considered exploratory.

Participant characteristics

A total of 414 women seeking fertility care were assessed for eligibility across the three participating hospitals. Twelve women (2.9%) met at least one exclusion criterion and were not enrolled. The remaining 402 women (97.1%) met the eligibility criteria and all provided written informed consent, giving an eligible-participant response rate of 100%. Each participant completed the questionnaire and provided an endocervical specimen for culture. No enrolled participant was excluded because of incomplete data or a missing laboratory result; therefore, all 402 participants, comprising 134 women from each hospital, were included in the final analysis (Fig 1).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Study flow chart showing participant screening, enrolment, specimen collection, and laboratory outcomes.

https://doi.org/10.1371/journal.pgph.0006796.g001

Most participants were aged 30–39 years (252/402, 62.7%), married (264/402, 65.7%), and had primary education (165/402, 41.0%). Informal employment was the most common occupation (177/402, 44.0%), and 178 women (44.3%) reported a monthly household income of less than 200,000 Ugandan shillings. Most participants reported no alcohol use (320/402, 79.6%), first sexual intercourse at 15–17 years (240/402, 59.7%), and no new sexual partner in the preceding 3 months (359/402, 89.3%). A total of 167 women (41.5%) reported three or more lifetime sexual partners, and 76 (18.9%) reported a history of vaginal douching.

A history of abnormal vaginal discharge was reported by 362 women (90.1%). Fourteen participants (3.5%) had a history of ectopic pregnancy. More than half had never used oral contraceptives (234/402, 58.2%), and 221 (55.0%) reported infertility duration of more than 3 years. Most participants were HIV-negative (354/402, 88.1%), 117 (29.1%) reported a history of anemia, and normal body mass index was the most common body mass index category (240/402, 59.7%). Sociodemographic, behavioral, obstetric, gynecologic, and medical characteristics did not differ significantly across the three study sites (Table 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of the study participants (N = 402).

https://doi.org/10.1371/journal.pgph.0006796.t001

Prevalence of gonococcal cervicitis

Of the 402 women seeking fertility care, 34 had culture-confirmed GC, giving an overall prevalence of 8.5% (95% CI, 6.1%–11.6%). The remaining 368 women (91.5%) were culture-negative (Fig 2).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Pie chart showing the overall prevalence of gonococcal cervicitis among women seeking fertility care (N = 402).

https://doi.org/10.1371/journal.pgph.0006796.g002

The prevalence varied modestly by study site. The Jinja Regional Referral Hospital had the highest prevalence at 10.4% (14/134; 95% CI, 6.3–16.9%), while the Lira Regional Referral Hospital and Kayunga Regional Referral Hospital each had a prevalence of 7.5% (10/134; 95% CI, 4.1–13.3%). These differences were not statistically significant (χ² [2] = 1.03, p = 0.598).

Antimicrobial susceptibility patterns of Neisseria gonorrhoeae isolates

Antimicrobial susceptibility results for the 34 culture-positive isolates are shown in Table 2.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Overall antimicrobial susceptibility patterns of Neisseria gonorrhoeae isolates (n = 34).

https://doi.org/10.1371/journal.pgph.0006796.t002

Among the 34 isolates tested, 32 (94.1%) met the CLSI susceptible-only disk-diffusion thresholds for ceftriaxone and azithromycin. Two isolates (5.9%) produced zone diameters below the susceptible-only threshold for each agent. Because MIC confirmation was not performed, these findings were reported as reduced zone diameters rather than intermediate or resistant results.

In contrast, resistance was frequent to penicillin, ciprofloxacin, and tetracycline. Resistance was observed in 22 isolates (64.7%) for penicillin and 20 isolates (58.8%) each for ciprofloxacin and tetracycline. Susceptibility was observed in 14 isolates (41.2%) for ciprofloxacin, 12 (35.3%) for tetracycline, and 11 (32.4%) for penicillin. Two tetracycline isolates (5.9%) and one penicillin isolate (2.9%) had intermediate results; no intermediate result was observed for ciprofloxacin.

Factors associated with gonococcal cervicitis

Bivariable analysis.

In the bivariable logistic regression, several variables met the prespecified screening threshold for inclusion in the multivariable model. These included age group, monthly household income, history of ectopic pregnancy, duration of infertility, body mass index category, number of lifetime sexual partners, history of vaginal douching, oral contraceptive use, HIV status, and reporting a new sexual partner in the previous three months (Table 3).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Bivariable logistic regression of factors associated with gonococcal cervicitis (selected variables shown).

https://doi.org/10.1371/journal.pgph.0006796.t003

Multivariable analysis.

In the multivariable logistic regression, five variables remained independently associated with GC (Table 4). Women reporting three or more lifetime sexual partners had significantly higher odds of gonococcal cervicitis than women reporting one partner (adjusted odds ratio [AOR], 5.25; 95% CI, 1.65–16.72; p = 0.005). A history of vaginal douching remained strongly associated with infection (AOR, 5.59; 95% CI, 1.83–17.03; p = 0.002). Current oral contraceptive use was independently associated with gonococcal cervicitis (AOR, 8.32; 95% CI, 2.94–23.54; p < 0.001), whereas past oral contraceptive use was not (AOR, 1.44; 95% CI, 0.43–4.87; p = 0.555). HIV-positive women had higher odds of gonococcal cervicitis than HIV-negative women (AOR, 5.00; 95% CI, 1.82–13.76; p = 0.002). Reporting a new sexual partner in the preceding 3 months was independently associated with infection (AOR, 7.92; 95% CI, 2.68–23.41; p < 0.001).

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Multivariable logistic regression of factors associated with gonococcal cervicitis.

https://doi.org/10.1371/journal.pgph.0006796.t004

Age group, monthly household income, history of ectopic pregnancy, duration of infertility, and body mass index category did not retain statistical significance after adjustment. The final multivariable logistic regression model showed no evidence of lack of fit on the Hosmer–Lemeshow goodness-of-fit test (χ² = 3.16, df = 8, p = 0.924). Neither of the prespecified multiplicative interactions was supported: HIV status × new sexual partner (LR χ²(1) = 0.17, p = 0.680) or vaginal douching × abnormal vaginal discharge (LR χ²(1) = 0.30, p = 0.584). The interaction terms were therefore not retained in the final model. Given the small number of culture-positive outcomes and the limited numbers in some joint-exposure categories, these analyses were considered exploratory.

Study strengths and limitations

This Multicenter study included three tertiary hospitals from geographically distinct regions of Uganda, strengthening its relevance to public fertility-care services. It assessed prevalence, antimicrobial susceptibility and associated factors within the same population, allowing the clinical, laboratory and epidemiological findings to be interpreted together. Culture-based diagnosis also provided viable isolates for phenotypic antimicrobial susceptibility testing rather than relying on symptoms or syndromic diagnosis alone.

Several limitations should be considered. The cross-sectional design prevents determination of whether the identified behavioural and clinical factors preceded infection or whether the detected infections contributed to infertility. Because recruitment was limited to women attending tertiary fertility clinics, the findings may not be generalizable to women who do not seek fertility care or who attend lower-level facilities. Culture is less sensitive than nucleic acid amplification testing, particularly when bacterial load is low or specimen transport and processing are suboptimal; therefore, the reported prevalence may underestimate the true burden of infection. Behavioural and reproductive information was self-reported and remains susceptible to recall and social-desirability bias despite the use of private interviews and neutral questioning.

The sample-size calculation did not account for possible clustering within hospitals. Although the achieved sample exceeded the individually calculated minimum and prevalence did not differ significantly across hospitals, residual hospital-level correlation cannot be excluded. In addition, only 34 participants had positive cultures, limiting the precision of the antimicrobial-susceptibility estimates and multivariable associations.

Antimicrobial susceptibility was assessed by disk diffusion without MIC confirmation. Consequently, isolates producing ceftriaxone or azithromycin inhibition zones below the susceptible-only thresholds could not be classified as confirmed resistant. Run-level quality-control measurements and documentation of repeat testing were also unavailable in the study dataset; therefore, quality-control discordance could not be assessed retrospectively. The small number of culture-positive outcomes limited the power to detect interaction; therefore, the absence of statistically significant interaction terms should not be interpreted as evidence that effect modification does not exist. Only 34 culture-positive outcomes were observed. Consequently, multivariable estimates remained vulnerable to sparse-data bias, limited precision, and model instability even after restricting the number of predictors and using penalized regression. The associated factors should therefore be interpreted as exploratory and require confirmation in larger studies.

Implications and future research

The coexistence of gonococcal infection, HIV and resistance to older antimicrobial agents highlights an opportunity to strengthen integration of fertility, STI and HIV services. Where feasible, fertility care could incorporate laboratory-based STI assessment, risk-reduction counselling, partner management and antimicrobial-resistance surveillance. However, the findings from this single cross-sectional study are insufficient to recommend universal routine screening.

Larger prospective studies should evaluate the feasibility and cost-effectiveness of gonococcal screening in fertility clinics, include sensitive molecular testing for N. gonorrhoeae and coexisting infections such as Chlamydia trachomatis, and investigate reproductive outcomes following diagnosis and treatment. Future antimicrobial-resistance studies should use validated MIC methods to confirm reduced ceftriaxone and azithromycin susceptibility.