**3. Results**

During recruitment, *n* = 1286 women were screened for study inclusion eligibility, of which *n* = 577 did not meet the inclusion criteria, and *n* = 229 declined to participate. No women declined to participate due to the requirement of a stool specimen collection. A total of 480 women were enrolled in the study, and *n* = 456 (95%) women provided a stool specimen at baseline (baseline blood collection, not stool collection, was required for enrollment). A total of *n* = 441 (92%) women remained in the study until completion at 12 weeks, with *n* = 434 (90%) women providing a 12-week stool specimen, depicted in Figure 3. Women who provided a baseline stool specimen (*n* = 456) and those that did not give a baseline stool specimen (*n* = 22) differed by education level achieved (fisher's exact, *p* = 0.044), breastfeeding status (fisher's exact, *p* = 0.007), reported diarrhea (3+ loose bowel movements in 24 h) (fisher's exact, *p* = 0.023), pain when passing stool (fisher's exact, *p*= 0.015), blood in stool (fisher's exact, *p* = 0.028), and if women had previously taken antibiotics (fisher's exact, *p* = 0.006) (Table 1).

**Figure 3.** Participant flow chart.

**Table 1.** Baseline participant characteristics of enrolled Cambodian women by provision of baseline stool specimen.


Total *n =* 480. Values are *n* (%) or median (IQR). 1 Independent samples *t*-test (parametric) and Wilcoxon rank sum tests (non-parametric) for continuous variables and Fisher's exact tests for categorical variables. \* Statistically significant at *p* < 0.05.

Throughout the study duration, research staff informally collected feedback from participants on their experience in the stool collection method. Some specific comments from research participants included: constipation made stool collection a challenge, this was their first time providing a stool specimen, and lastly, in general, participants expressed greater

hesitation and fear towards blood collection than stool collection. It should be emphasized that this data was not systematically collected and may be biased by many factors, such as response bias. Research staff shared that verbal communication was more productive and helpful to participants than the written instructions (Khmer translated infographic).

DNA was extracted from a subsample (*n* = 150) of thawed neat stool using QIAamp PowerFecal Pro DNA Kit [10], and DNA yield was checked via NanoDrop spectrophotometer reading. All extracted specimens yielded DNA suitable for PCR amplification and were thus uncompromised during specimen collection, transportation, and storage.
