This study was designed as a two-armed prospective pilot study, with an intervention and a control group and was conducted to inform the future design of a prospective, randomized controlled trial [11]. The study was conducted in a gynecological clinic in a community hospital located in southwest Sweden. Patients are referred to the hospital from primary care clinics, maternal health clinics or other hospitals in the region, and may also request care from the clinic by self-referral, which is then evaluated and responded to by the gynecological team.

The study was registered in the Research Database Registry in Västra Götalandsregionen (Sweden) with number: 254161 on 10 November 2018.

Patients who attended the clinic between October 2018 to April 2020 were eligible for participation in the study if they were at least 18 years old and had one of the following indications of office hysteroscopy: abnormal pre- or postmenopausal uterine bleeding; endometrial thickening or polyps; submucosal and some intramural fibroids; intrauterine adhesions; uterine malformations; retained intrauterine contraceptives or removal of foreign bodies. The patients also needed to understand written Swedish to apprehend the study information and consent form and respond to the self-rated questionnaires. Exclusion criteria were viable intrauterine pregnancy, active pelvic infection (including genital herpes infection) or known cervical or uterine cancer, and poor ability to understand Swedish.

Eligible patients were informed, verbally, and in print, about the study by their treating gynecologist, before their office hysteroscopy appointment. Each patient was given time to ask questions and consider the invitation to take part at the following visit. Participants who accepted the invitation to participate signed a consent form and were randomized to one of two groups, the doula intervention or routine care.

Participants who accepted the invitation to participate signed a consent form and were allocated to either the doula intervention or routine care. A simple randomization technique, based on week numbers (even/odd) of enrolment, was used to assign the patients to either of the two groups (Doula/control). Although simple randomization increases the risk of imbalanced groups (both numbers and covariates) [12], it was chosen for pragmatic reasons related to resources and the clinical setting. The randomization and assignment was conducted by a medical secretary who was not involved in the study. To minimize bias based on patient expectations, allocation was not revealed to the patients (before the office hysteroscopy). They were informed about the scheduled time and place and received general information about the gynaecological procedure.

The office hysteroscopy procedure and materials used during the procedure were identical in both groups, with the same experienced gynecologist treating all patients. The office hysteroscopy was conducted outside of the operating theater setting in an appropriately sized, equipped, and staffed treatment room with adjoining, private changing facilities and toilets. The standard appointment timeframe was 45 min. In the line with local clinical practice guidelines for office hysteroscopy the women were instructed to take standard doses of NSAIDs (400 mg Ibuprofen) around 1 hour before their scheduled appointment to reducing pain during and after the procedure.

Miniature hysteroscopies (CAMPO TROPHYSCOPE® with Sheath with gliding mechanism for a primary approach to uterine cavity only with 2.9 mm outer diameter and a working channel allows the use of semirigid 5 Fr. operating instruments and bipolar electrodes) were used. All patients received local anesthesia with a paracervical block of 6 mL of mepivacaine 10 mg/mL in the cervix.

In line with routine office hysteroscopy, the choice of distension was normal saline [13] and the pressure of the distension medium was 70 mmHg. With the patient in a dorsal lithotomy position, a vaginoscopy was performed, which is the standard technique. The hysteroscopy was connected to a camera and a monitor. Depending on the findings, the following procedures were performed as required: biopsies, polypectomies, or myomectomies with the same hysteroscope. In case of failed hysteroscopy or other medical reason, for example findings that required surgery, patients were referred to operative hysteroscopy under general anesthesia at the same hospital.

A doula accompanied the participants throughout the procedure, starting from the moment the patient registered at the front desk of the hospital and ending when they left the clinic. In line with documented elements of doula support in the childbirth context [14], the doula in the present study aimed to provide: (a) attention to physical comfort; (b) emotional support (praise, reassurance, encouragement, and continuous presence); (c) information sharing (nonmedical advice, explanation of policies and procedures, anticipatory guidance) and (d) advocacy (facilitation of communication between the woman and hospital staff to assist in making informed decisions). The doula was qualified, had 12 years of experience as a doula, and was employed as an assistant nurse at the clinic. To avoid spill-over effects, she was not attending to patients in the control group.

The control group received routine care, which meant that the patients, after registration, were seated in the gynecology clinic waiting room. They received basic support from an assistant nurse during the office hysteroscopy and were provided a comfortable and safe environment with acoustic, thermal, and visual comfort.

Sample size: A total of 28 women were included, 12 in the Doula intervention group and 16 in the routine care group.

The sample size was planned in line with the rule of thumb for pilot studies [15], suggesting 12 subjects in each group. The justification of this rule of thumb was based on a rationale about feasibility, and precision about mean and variance. Descriptive statistics were used to describe the clinical and background variables. For the continuous variables pain (NRS) and anxiety (STAI-S), differences between groups were analyzed using analysis of variance ANOVAs (continuous variables) and Fisher’s exact test (categorized data).

Data for the STAI-S did not violate assumptions of normality according to the Shapiro Wilk’s test. The dichotomized pain variable (NRS) was analyzed using the chi-square test. A p-value of $\le$0.05 was set as the limit for statistical significance in all tests. IBM SPSS Statistics version 25.0 (IBM Corp, Armonk, NY, USA) was used for all statistical analyses.

Background data were collected by the gynecologist or a research assistant at the start of the appointment, including age, body mass index, the number of vaginal deliveries, educational achievement level, and history of endometrial surgery (curettage and/or hysteroscopy), diabetes, age, previous curettage, dyspareunia, severe dysmenorrhea, and previous hysteroscopy experience. Cervical stenosis was classified based on localization: stenosis of external cervical ostium (ECO; type I); stenosis of the distal third of cervical channel and the internal cervical ostium (ICO; type II); stenosis of the ICO (type III), and combined stenosis of ECO and ICO (type IV) [16]. Outcome data were collected at the start of the office hysteroscopy, and separate phases during the procedure.

Feasibility was evaluated by three outcomes:

(1) Number of successful (access to and visualization of the entire uterine cavity was possible during the same procedure), partially successful (access to uterine cavity possible, but the entire uterine cavity could not be examined) or failed (access to uterine cavity was not possible) hysteroscopies.

(2) Duration of the procedure.

(3) Complications (bleedings, false tracks, uterine perforation or cervical tears, or reactions to mepivacain (numbness and tingling of the lips or tongue, nervousness, dizziness, blurred vision, tremors, drowsiness, convulsions, sweating, unconsciousness or respiratory arrest)).

The Spielberg State-Trait Anxiety Inventory (STAI) was used to measure the patients’ state of anxiety before and after the examination [2, 17, 18]. The STAI is a widely used measure of anxiety, validated among perinatal women, where it has shown acceptable sensitivity, specificity, and predictive values to identify cases of anxiety [19]. The assessment consists of two subscales, the STAI-S which measures the individual’s anxiety in the present situation and the STAI-T which measures anxiety as a general trait in the individual. Each subscale has 20 items with each item rated using a 4-point Likert scale and a score $>$40 as the suggested cutoff for clinically significant anxiety symptoms [20]. For this study, to minimize the time and number of questions asked to the participants, only the STAI-S was used to measure the situational anxiety pre- and post-procedure.

The Numeric Rating Scale (NRS), an instrument for self-reported pain intensity, was used during the procedure. The subject reports a number between 0 and 10, where 10 stands for worst possible pain, 0 indicates no pain. The NRS is similar to the commonly used Visual Analog Scale (VAS) but is less influenced by non-pain intensity factors [21]. Pain intensity was assessed via NRS 0–10 at four points in time during the procedure:

1. Introduction of hysteroscopy into the vagina.

2. Progression through the cervical canal up to the internal uterine orifice.

3. Inspection of the uterine cavity.

4. Performing the endometrial biopsy/polypectomy etc. if required.

Previous well-defined limits for unacceptable pain were used [22]. Pain was considered unacceptable if severe pain was reported during the procedure (NRS $\ge$7). In these cases, the office hysteroscopy was discontinued. The pain variable was also dichotomized, defined as $>$4 = moderate-to-severe pain [23].

In two cases (n = 2), one in the doula group and one in the control group, office hysteroscopy was classified as non-successful. In the doula group, one office hysteroscopy could not be performed due to pain at the entrance to the cervix, so a blind endometrial biopsy was performed. In the other case, the polyp recession could be started but had to be stopped due to pain and bleeding. The first case was scheduled for a hysteroscopy under general anesthesia, whereas in the second case, a second office hysteroscopy was performed, and the polyp was completely resected.

In the doula group, three hysteroscopies were classified as partially successful. In these cases, biopsies were performed, and the base of the polyp was sectioned, even though the polyp could not be fully removed. After three months, ultrasound controls were normal and did not require further interventions.

There were no significant differences between the groups regarding cervical stenosis. Cervical stenos type II was found in one woman from each group, and one cervical stenosis type III was found in the control group.

The mean duration of the procedure was 18.6 minutes (range 4–40) in the doula group and 18.9 minutes (range 8–40) in the control group (Table 2), with no significant difference between groups. There were no procedure-related complications (bleedings, false tracks, uterine perforation or cervical tears) in any of the groups, and no participant in any of the two groups reported side-effects of mepivacain.

There were no differences in pain assessments between the groups, see Table 2. For the anxiety outcome STAI-S, both groups had high, comparable levels of state anxiety before the procedure. After the procedure, the doula group showed slightly increased state anxiety, while the control group showed slightly decreased anxiety. There was a significant difference between groups favoring the control group when comparing STAI-S mean score post-procedure, but not when analyzing mean change across groups (Table 2 and Fig. 1). There were no significant differences in post-procedural anxiety and age, duration of the procedure, the reason for office hysteroscopy, or parity.

The present pilot study showed a mean pain score of VAS 35 mm during office hysteroscopy, with a mean standard deviation of 18 mm for the whole sample. In line with a previous RCT [24], ewe considered a 10 mm difference in mean pain between the two groups to be clinically relevant. Based on these figures, we conducted a post hoc power analysis which showed that, to detect a minimal 10 mm difference between the groups, with 80% power on the 0.05 significance level, 51 patients would be required in each group. Assuming 20% dropout [24], we suggest that a future full scale RCT based on this pilot would need 61 patients in each group, resulting in a total required sample of 122 patients.