Vaginal dinoprostone versus placebo for pain relief during intrauterine device insertion: a systematic review and meta-analysis of randomised controlled trials

Abstract Objective To systematically and meta-analytically investigate the efficacy and safety of vaginal dinoprostone versus placebo in pain relief during intrauterine device (IUD) insertion. Methods PubMed, Scopus, Web of Science, and Cochrane Library were screened till 1 October 2020. Only randomised placebo-controlled studies were included and assessed for risk of bias. Main outcomes included IUD insertion related pain, patient satisfaction, provider ease of IUD insertion, and side effects. Pooled outcomes were summarised as standardised mean difference (SMD), weighted mean difference (WMD), or risk ratio (RR) with 95% confidence interval (95% CI). Results Five studies were included, comprising 862 patients; equally 431 patients received vaginal dinoprostone and placebo. All studies showed an overall low risk of bias. When compared to placebo, dinoprostone significantly correlated with decreased pain at tenaculum placement (SMD = −0.79, 95% CI [−1.43, −0.16], p = 0.01), decreased pain at uterine sounding (SMD = −0.88, 95% CI [−1.54, −0.22], p = 0.009), decreased pain at IUD insertion (SMD = −1.18, 95% CI [−1.74, −0.61], p < 0.001), decreased need for additional analgesia (RR = 0.34, 95% CI [0.22, 0.53], p < 0.001), increased patient satisfaction (SMD = 1.41, 95% CI [0.62, 2.20], p < 0.001), and increased provider ease of IUD insertion (SMD = −1.17, 95% CI [−1.62, −0.73], p < 0.001). Fever was statistically significantly higher in dinoprostone versus placebo group (RR = 3.73, 95% CI [1.47, 9.44], p = 0.006). All other side effects—including nausea, vomiting, shivering, diarrhoea, abdominal cramps, vasovagal attack, uterine perforation, and postprocedural bleeding—did not substantially differ between both groups. Conclusions This first ever meta-analysis advocates that dinoprostone compared with placebo is safe, effective, and yields favourable analgesic outcomes during IUD insertion.


Introduction
Intrauterine devices (IUDs) deliver a largely effective, usually safe, and long-lasting method of reversible contraception with an analogous efficacy to tubal sterilisation [1]. The use of IUDs is globally endorsed by the American College of Obstetricians and Gynaecologists [2], Society of Obstetricians and Gynaecologists of Canada [3], and European Society of Contraception [4] to avoid unplanned pregnancies among sexually active adolescents and young women. Globally, the two most frequently utilised IUDs comprise the levonorgestrel-releasing intrauterine system (LNG-IUS) and copper-containing intrauterine device (Cu-IUD) [1,2], both of which are equally used in nulliparous and multiparous women [5].
There are a few downsides associated with IUD use. Importantly, pain perception is one of the substantial factors contributing to restricted utility of IUDs, particularly among nulliparous women who possess relatively narrower uterine cavities when compared to their multiparous counterparts [6,7]. This remark is in line with the perspective that a large proportion of healthcare personnel restrict IUD administration to nulliparous women owing to worries pertaining to anticipated insertion pain and procedural difficulties [8][9][10]. Indeed, each step of IUD insertion procedure can instigate a large deal of pain perception [11]. Therefore, proper control of pain during IUD insertion procedure is critically important to allow for optimising a patient's access and choice of contraceptive method.
The optimal method of pain relief during IUD insertion remains undefined [12,13]. A contemporary systematic review and network meta-analysis of numerous lines of pharmacologic analgesic interventions-including placebo, nonsteroidal anti-inflammatory drugs, nitric oxide donors, lavender scent, lidocaine, and misoprostol-demonstrated no tangible effectiveness for IUD insertion-related pain [14]. Conversely, lidocaine-prilocaine cream (genital mucosal application) was the most effective intervention for IUD insertion-related pain [14]. However, when lidocaine-prilocaine was compared head-to-head with other pharmacologic interventions, including placebo, it did not exhibit significantly reduced pain 5-20 min after IUD insertion [14]. More research is warranted for alternative analgesics to control pain during IUD insertion procedure.
Dinoprostone is a naturally occurring prostaglandin E2 equivalent [15]. It is frequently employed in obstetrics for labour induction, which is mediated through cervical ripening and prompting of uterine contractions with an equivalent labour-inducing efficacy to misoprostol but substantially fewer adverse events [16]. Additionally, dinoprostone has been exploited successfully prior to diagnostic hysteroscopy to ease the procedure and reduce associated pain without considerable toxicity when compared to placebo or misoprostol [17,18]. Only a very limited number of trials examined the safety and efficacy of vaginal dinoprostone versus placebo in facilitating IUD insertion and decreasing its related pain [12,[18][19][20][21]. To date, no meta-analysis has been conducted to amass the data and inform concrete conclusions. Therefore, the aim of this study is to systematically and meta-analytically synthesise evidence from randomised controlled trials that scrutinised the safety and efficacy of vaginal dinoprostone versus placebo for pain relief among women undergoing IUD insertion.

Methods
This systematic review and meta-analysis was conducted in harmony with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [22].

Literature search strategy
Four databases (PubMed, Scopus, Web of Science, and Cochrane Library) were screened from inception to 1 October 2020. The following search strategy was used for all databases: (dinoprostone OR cervidil OR prepidil) AND (intrauterine device OR iud). There was no language restriction.

Inclusion and exclusion criteria
We included all articles that met the following criteria for our PICOS evidence-based research question: (I) Patients: women who received LNG-IUS or Cu-IUD for contraception, (II) Intervention: vaginal dinoprostone, (III) Comparator: vaginal placebo, (IV) Outcomes: efficacy and safety endpoints, and (V) Study design: randomised controlled trials.
We excluded drugs other than dinoprostone, indications other than contraception, non-randomised study designs, non-human trials, abstracts, and articles without full texts.

Screening of results
The retrieved citations were exported using EndNote software and duplicates were crossed out. The screening of results was completed through a two-fold phase. The first phase involved title and abstract screening of all citations. The second phase involved retrieval of the full text of all potential citations. Moreover, the reference lists of the included studies were reviewed manually for potential inclusion of other relevant studies. Two authors screened the citations independently and disagreements were resolved by a consensus among the two authors.

Risk of bias assessment of the included studies
We utilised the Cochrane's risk of bias tool in evaluating the quality of the included studies [23]. This tool is elaborated in the Cochrane Handbook for Systematic Reviews of Interventions, Version 5.1.0, Chapter 8. This tool appraises the following domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, and other potential sources of bias. Each domain was judged as 'low', 'high', or 'unclear' risk. Two authors evaluated the risk of bias independently and disagreements were resolved by a consensus among the two authors. An overall high risk study was established if the study had at least two or more domains that were scored as 'high' risk.

Data extraction
We extracted three types of data: (I) baseline characteristics of the included studies, (II) efficacy outcomes, and (III) safety outcomes. Data about baseline characteristics of the included studies comprised author's first name, year of publication, national clinical trial identifier, country, type of IUD device, study group, sample size, timing of drug administration before IUD procedure, and the person administering the drug. Moreover, patients' age, body mass index, parity, gravidity, position of uterus, history of previous abortion, and history of previous IUD were extracted. Data about efficacy outcomes included pain at tenaculum placement, pain at uterine sounding, pain at IUD insertion, pain after IUD insertion (10-30 min), need for additional analgesia after procedure, duration of procedure, ease of IUD insertion as reported by healthcare providers, and procedural satisfaction as reported by patients. Pain scores were evaluated according to the 10-cm/100-mm visual analogue scale (VAS) in which '0' corresponded to no pain at all and '10 cm/100 mm' corresponded to the worst possible pain imaginable. Likewise, ease of IUD insertion as reported by healthcare providers was scored according to a 10-cm/ 100-mm VAS-like metric in which '0' corresponded to easy insertion and '10 cm/100 mm' corresponded to extremely difficult insertion. Equally, procedural satisfaction as reported by patients was scored according to a 10-cm/100mm VAS-like metric in which '0' corresponded to no satisfaction and '10 cm/100 mm' corresponded to maximum satisfaction. Data about safety outcomes included nausea, vomiting, diarrhoea, shivering, fever, abdominal cramps, postprocedural bleeding, vasovagal attack, and uterine perforation. Six authors participated in data extraction and verification.

Data analysis
Review Manager Software version 5.4 was used for metaanalysis. Continuous data were analysed using the inverse variance method and reported as weighted mean difference (WMD) or standardised mean difference (SMD), as appropriate, with 95% confidence interval (95% CI). Dichotomous data were analysed using the Mantel-Haenszel method and reported as risk ratio (RR) with 95% CI. Statistical heterogeneity was established if chi-square was p < 0.1 and I-square test (I 2 ) was >50% [24]. Fixed-and random-effects models were used for meta-analysis of homogeneous and heterogeneous data, respectively. Publication bias was not evaluated since the number of included studies (n ¼ 5) was lower than the minimum required (n ¼ 10) [25].

Search results and summary of included studies
Literature search generated a total of 67 studies after omission of duplicated ones. After title and abstract screening, 50 studies were excluded and the remaining 17 studies progressed to full text screening for eligibility. Finally, a total of five (n ¼ 5) studies met the inclusion criteria and were included in the qualitative and quantitative synthesis [12,13,[19][20][21]. Figure 1 displays the PRIMSA flowchart. This meta-analysis included 862 patients; equally 431 patients received vaginal dinoprostone and placebo. All studies originated in Egypt. Three studies included nulliparous women, whereas one study included only patients who delivered by caesarean section. Furthermore, three studies included patients receiving Cu-IUD as the method of contraception. The dose of vaginal dinoprostone was consistent (3 mg) in all studies, however, the duration of drug application differed between studies, ranging from 2 to 12 h before procedure. Drugs were administered by nurses in three studies and self-administered in two studies. The baseline characteristics of the included studies are displayed in Table 1.

Risk of bias assessment of the included studies
Overall, two studies [13,19] had high risk of bias on one element, but all studies were considered to have low risk of bias. In two studies [13,19], the drugs (vaginal dinoprostone and placebo) were self-administered by patients. Although measures had been taken by investigators to remind patients about the time to vaginally self-administer the drugs three [19] and 12 [13] hours before the procedure, however, this cannot be certainly established, and we judged the other bias domain as 'high' risk. The graph and summary of risk of bias are depicted in Figure 2.

Findings and interpretation
This systematic review and meta-analysis endeavoured to examine the safety and efficacy of vaginal dinoprostone versus placebo in controlling pain during IUD insertion. We included five low risk of bias randomised controlled trials comprising 862 patients (dinoprostone, n ¼ 431 and placebo, n ¼ 431).
Despite IUD insertion is relatively a quick (5-10 min) procedure, each step of it can bring about variable extents of pain [11]. Sources of pain comprise speculum insertion, tenaculum placement, transcervical sounding of uterus, forward advancement of IUD inserter into uterine cavity, and postprocedural pain. When compared to placebo, our pooled analyses showed that dinoprostone significantly reduced pain at tenaculum placement, transcervical sounding of uterus, and IUD insertion. This statistically significant pain reduction was positively correlated with increased patient-reported procedural satisfaction and decreased requirement for additional postprocedural analgesia. However, it must be noted that this pain reduction was not clinically meaningful. The minimum clinically meaningful difference in VAS pain score was previously delineated by Todd and colleagues [26] as the numeric change in VAS pain score that is associated with a patient's subjective assessment of a little less pain or a little more pain. According to family planning literature, a reduction of 1.5cm difference is regarded as clinically meaningful. On the other hand, Rowbotham et al. [27] suggested a minimum decrease of 2 points or 30% in order to attain a clinically meaningful reduction in pain. By applying both cut-offs, while pain reduction with dinoprostone versus placebo was statistically significant, it was nevertheless not clinically meaningful.
Additionally, our summary analysis showed that although pain after IUD insertion was reduced, this pain reduction did not reach statistical significance. The lack of statistically significant pain reduction after IUD insertion could be credited to the variable time points that were used to assess this parameter across all the pooled four studies (10, 15, 20, and 30 min). Thus, this observation might have introduced a bias in the assessment of postprocedural pain of the pooled outcome.
Our findings indicated that dinoprostone was correlated with significantly increased ease of IUD insertion by healthcare providers. Nonetheless, the IUD insertion time was not substantially impacted. This increased ease of insertion score could be ascribed to the favourable cervical ripening effects of dinoprostone [15,28]. Mechanistically, dinoprostone initiates cervical softening through stimulation of interleukin 8 (IL-8), which in turn facilitates influx of polymorphonuclear leukocyte neutrophils that orchestrate remodelling of cervical extracellular matrix and induction of progesterone withdrawal [28].
There are some risk factors that may produce more severe pain perception during IUD insertion. Such factors comprise nulliparity, women who delivered only by caesarean section, and use of LNG-IUS [29][30][31]. When compared to parous women, nulliparous women relatively possess narrower uterine cavities which may not properly fit the dimensions of conventional IUDs, thus culminating in higher IUD insertion pain [7]. Moreover, while a history of caesarean section does not automatically preclude IUD insertion, nevertheless, a structurally disfigured uterus secondary to repetitive caesarean section may potentially correlate with difficult IUD insertion and discomfort [32]. Lastly, the comparatively thicker diameter of LNG-IUS inserter (ranging from 4.65 to 4.85 mm) contrasted to Cu-IUD counterpart (around 4.0 mm) may cause more pain during insertion [31]. In our study, subgroup analyses according to parity and device of IUD revealed that dinoprostone equally significantly correlated with better pain control, greater ease of IUD insertion, and higher patient satisfaction during the procedure when compared to placebo (Supplementary File 2). This study suggests that dinoprostone can yield successful insertion of IUD irrespective of the parity status and type of IUD with favourable efficacy outcomes.
Our summary data depicted that side effects did not significantly differ between dinoprostone and placebo groups. Fever was the only drug-related side effect that was significantly higher in the dinoprostone group when compared to placebo group. This side effect is, to a larger degree, expected as the association between prostaglandin E2 and occurrence of fever is well documented in literature [33]. Other than fever which can be adequately and conservatively managed by antipyretics, our study suggests  (77) 17 (17) BMI: body mass index; Cu-IUD: copper-containing intrauterine device; LNG-IUS: levonorgestrel-releasing intrauterine system, NCT: national clinical trial identifier (clinicaltrials.gov); NR: not reported; Ref: reference; SD: standard deviation.
dinoprostone is highly effective and associated with satisfactory safety profile.

Relevance of findings: implications for clinical practice
Dinoprostone is widely used for the obstetric indication of labour induction owing to favourable cervical softening and uterine contractility properties [15,28]. A meta-analysis demonstrated that while dinoprostone is efficaciously equivalent to misoprostol in labour induction, it is associated with better toxicity profile, particularly in terms of lower frequencies of tachysystole and hypertonic uterine dysfunction [16]. Also, use of dinoprostone has been extended to non-obstetric indications such as hysteroscopy in nulliparous [17] and postmenopausal [18] women and correlated with better pain control and ease of procedure when compared to placebo or misoprostol. This meta-analysis further supports the clinical utility of dinoprostone in an additional non-obstetric indication-that is, during IUD insertion to control procedure-related pain and facilitate its ease of conduction. Dinoprostone possesses two major drawbacks that ought to be acknowledged. First, from a financial perspective, dinoprostone is costly when compared to its closely related comparator misoprostol [34,35]. Thus, there is a need to judiciously identify those women who may benefit clinically from dinoprostone. Second, from a physiochemical perspective, dinoprostone is unstable at room temperature and should be stored in freezer/refrigerator until before use to preserve its potency [12].
Accumulating body of evidence suggests that changing the IUD insertion technique before analgesic administration may help to reduce pain during the procedure. To elaborate, it has been shown that a tenaculum is not always needed [36]. Furthermore, it has been demonstrated that uterine sounding is unnecessary [37] and inaccurate when compared with hysteroscopy [38]. The 'direct method' of IUD insertion can replace the standard method. The 'direct method' generally aims to insert the IUD device without the use of a tenaculum or uterine sounding. Bastin et al. [36] recently compared pain outcomes in women undergoing the direct versus standard method of IUD insertion. The authors found that the 'direct method' of IUD insertion correlated with significantly reduced pain and higher patient satisfaction without an increase in the rate of adverse events [36]. Thus, ritualising the IUD insertion procedure, possibly through the application of the 'direct method' and effective analgesics (for example, dinoprostone), could substantially improve women's comfort and result in a greater acceptance of the contraceptive IUD.

Unanswered questions and future research
The optimal method of pain relief during IUD insertion remains undefined [12,13]. A contemporary systematic review and network meta-analysis of numerous lines of pharmacologic analgesic interventions showed that lidocaine-prilocaine cream (genital mucosal application) was the most effective intervention for controlling IUD insertion-related pain [14]. Dinoprostone was not included in the aforementioned network meta-analysis [14]. Thus, future research should be geared towards direct comparison of efficacy and safety of vaginal dinoprostone versus lidocaine-prilocaine cream (and other active comparators such as misoprostol, lidocaine, and nonsteroidal antiinflammatory drugs) in controlling pain associated with IUD insertion. This should be achieved through development of well-designed randomised controlled trials that take into account women who are at high risk for more painful experiences, such as nulliparous women, women who delivered only by caesarean section, women who failed previous insertions, and women who will receive the comparatively thicker LNG-IUS as opposed to Cu-IUD. Supplementary File 3 depicts a list of registered clinical trials (clinicaltrials.gov) that will examine the safety and efficacy of dinoprostone versus placebo or an active comparator (misoprostol, lidocaine-prilocaine, or isonicotinic acid hydrazide) in women undergoing IUD insertion.

Differences and similarities in relation to other studies
Comparative effectiveness studies between dinoprostone and active comparators are very limited. Our systematic literature search identified only one study that compared dinoprostone with misoprostol in nulliparous women undergoing IUD insertion [12]. The study found that there were no statistically significant differences between both groups with regard to operator-reported ease of insertion scores as well as patient-related pain scores (at tenaculum insertion, at uterine sounding, during IUD insertion, and 10 min after IUD insertion). Compared to placebo, both dinoprostone and misoprostol substantially reduced pain perception during IUD insertion procedure with statistical significance. Nevertheless, compared with placebo, the reduction in pain scores was clinically meaningful only in women who received dinoprostone.

Strengths and weaknesses
This study has several strengths. First, this is the first systematic review and meta-analysis that pooled the efficacy and safety of vaginal dinoprostone versus placebo in controlling pain during IUD insertion. We included only randomised placebo-controlled trials (n ¼ 5) to lessen potentials of bias and confounding in our pooled conclusions. All included studies were of low risk of bias. We strictly adhered to PRISMA guidelines during the conduction of this research. Moreover, we reported many efficacy and safety endpoints. All in all, our study suggests the beneficial role of vaginal dinoprostone for a non-obstetric indication, which is controlling pain during IUD insertion.
Nonetheless, this study has some limitations that should be recognised. Such limitations comprise the subjective evaluation of pain which may be impacted by the patients' sociodemographics or pre-anxiety levels. The studies varied with regard to sociodemographics (for example, parity, history of caesarean section, history of IUD use, and first time use of IUD), standardisation of IUD insertion procedure, timing of vaginal dinoprostone administration ranging from 2 to 12 h prior to IUD insertion, as well as timing of assessing pain after IUD insertion ranging from 10 to 30 min. All these variations, along with the type of drugadministering health care professional, could have negatively impacted the factual assessment of the efficacy endpoints. It should be noted that the optimal timing of dinoprostone administration is yet to be determined. Lastly, all the included studies originated from one country (Egypt) and the reason behind this observation remains unclear.

Conclusions
This systematic review and meta-analysis suggests that vaginal dinoprostone compared with placebo is correlated with increased ease of insertion by providers, higher satisfaction by patients, and decreased IUD insertion-related pain. Moreover, dinoprostone is largely safe with very tolerable toxicity profile. All in all, this meta-analysis supports the clinical utility of vaginal dinoprostone for pain relief during IUD insertion, including those nulliparas who are at high risk for increased pain perception.