Uterine perforation on intrauterine device insertion: is the incidence higher than previously reported?

1. Introduction 

Perforation of the uterus is a serious, if uncommon, complication of intrauterine device (IUD) insertion. Clinical trials of Multiload Cu375 have reported an incidence of less than one uterine perforation per 1,000 insertions [1], [2], [3]. Edelman’s review of Multiload Cu375 studies found no reported uterine perforations in 5617 insertions [4].

Studies of Multiload Cu375 before this post-marketing study have taken place in expert centers [1], [3] or with a limited number of inserting doctors [2] and with defined exclusion criteria [1], [2], [3]. The rate of perforation might be different in ‘real-life’ clinical practice outside the setting of a clinical trial, where many different doctors (of varying experience) insert IUDs in a range of women.

In New Zealand, Multiload Cu375 is included in the Intensive Medicines Monitoring Programme (IMMP) which undertakes prospective observational cohort studies on selected medicines in the early post-marketing period [5]. The IMMP has collected data from over 17,400 Multiload Cu375 insertions throughout New Zealand. This paper reports on the incidence of uterine perforation in this cohort and discusses why the incidence of this complication might differ between studies.

2. Materials and methods 

The New Zealand IMMP uses Prescription Event Monitoring (PEM) methodology to prospectively study medicines as used in clinical practice after licensing and has been described in detail previously [5], [6]. Multiload Cu375 has been included in the IMMP since it was first marketed in New Zealand in 1991.

Between 1991 and 1997, Multiload Cu375 packages included a registration form which was completed by the inserting doctor and returned to the IMMP in a pre-paid envelope. The registration form included personal details for each woman (including age and parity), details of the inserting doctor and place of insertion. Doctors were also asked to record any problems with insertion. Follow-up questionnaires were sent to the inserting doctors annually from approximately one year after insertion. The questionnaires requested information on all adverse events occurring since insertion of Multiload Cu375. Additional information was also received in the form of spontaneous reports as the IMMP is linked with the national spontaneous reporting program for adverse reactions.

Information from the registration forms, follow-up questionnaires and spontaneous reports was entered into the Multiload Cu375 database. A physician at the IMMP assessed adverse events reported (at any time). Events were coded using terms from the IMMP adverse events dictionary which is based on the World Health Organization Adverse Reactions Terminology (WHOART). The IMMP dictionary has additional terms for events related to IUDs and has an additional System Organ Class (SOC) ‘Insertion Problems’ to group together events related to insertion.

The Multiload Cu375 cohort was searched for all events that indicated perforation of the uterus had occurred. In addition to the term ‘uterus perforated,’ other terms searched for were ‘IUD intra-abdominal’ and ‘IUD embedded’. The search was not restricted to ‘Insertion Problems’ so that all perforations diagnosed at a later date (and coded in other SOCs) would be included. The two clinical investigators (MHW and DC) assessed each case report identified.

The total number of insertions reported by each doctor was analyzed and these were grouped into those who reported inserting less than 10 Multiload Cu375 devices, between 10 and 50 devices, between 50 and 100 devices, and more than 100 devices in the study period. The incidence of perforation events in each of these groups was determined and relative risks were calculated with 95% confidence intervals. All analyses of the IMMP database were performed using SAS software version 6.12.

3. Results 

The Multiload Cu375 cohort included 17,469 insertions (in 16,159 women) which took place between January 1991 and March 2001. The average age was 32 years (range 14 to 56 years) and for women whose parity was known (14,310 insertions), 40% had delivered two babies. About 9% of insertions were carried out in nulliparous women.

In this study, 1,699 different doctors in general practices, family planning clinics, hospitals and private specialist rooms throughout New Zealand performed the Multiload Cu375 insertions. General practitioners performed 92% of all insertions and the number of reported insertions by each individual doctor ranged from one to 293 during the study period.

3.1. Incidence of uterine perforation 

In total, 28 reports that suggested perforation of the uterus were identified, giving a total incidence of 1.6 per 1,000 insertions. The frequency and incidence of these ‘perforation events’ are shown in Table 1.

Table 1. Frequency and incidence of perforation events

	Event (IMMP term)	Frequency of event*	Incidence per 1000 insertions
	Uterus perforated	13	0.7
	IUD embedded	14	0.8
	IUD intra-abdominal	1	0.06
	Total	28	1.56

* All these events occurred in different women.

In total, 24 of the 28 (86%) perforation events were not diagnosed at the time of insertion. Table 2 shows the time from insertion of the device to diagnosis of the perforation event for all 28 cases identified. Nearly half of the cases (46%) were not identified for more than 1 year after insertion.

Table 2. Time from insertion of Multiload Cu375 to diagnosis of perforation events*

	Time from insertion to diagnosis of event	Number of perforation events	% all perforation events
	At time of insertion	4	14.3
	Less than 1 week	3	10.7
	1 week to 3 months	4	14.3
	4 months to 1 year	3	10.7
	1 to 2 years	7	25.0
	More than 2 years	6	21.4
	Unknown	1	3.6
	Total	28	100

* Perforation events included all events listed in Table 1.

Of the 13 events assessed as ‘uterus perforated,’ review of the reports showed that eight were not recognized at insertion. In these women, the time at which the perforation was diagnosed ranged from 1 day to 4 years after insertion. Seven of the 13 ‘uterus perforated’ reports stated that the IUD was found in the abdominal cavity, implying the device had completely perforated the myometrium. In one case, the uterus was perforated with the uterine sound before the device was inserted.

Of the 14 reports assessed as ‘IUD embedded,’ two women whose device was removed 4 months after insertion had a clinical history strongly suggestive of (unrecognized) perforation at the time of insertion. In another nine reports, the device was found embedded in the uterine wall several months to over 6 years later. The remaining three reports were of Multiload Cu375 devices found embedded after unintentional pregnancies with the device in situ.

The three different IMMP event terms (uterus perforated, IUD embedded and IUD intra-abdominal) reported were treated as one group termed ‘perforation events’ for subsequent analyses. Parity was reported for 23 of the 28 women in whom perforation events were reported. These 23 women had all delivered at least one baby with parity = 1 for five women, parity = 2 for 11 women and parity = 3 or 4 for seven women. Although the registration forms did not specifically request information about timing of IUD insertion relative to parturition, one report of uterine perforation stated the woman was seven weeks post-partum at the time of insertion. Another report of ‘IUD embedded’ occurred in a woman who was (unknowingly) 5 weeks pregnant at the time the IUD was fitted.

3.2. Experience of inserting doctor 

Of the 1699 different doctors who reported a total of 17,469 insertions, 1244 (73%) doctors reported fitting less than 10 Multiload Cu375 devices over the study period, 400 (24%) reported between 10 and 50 insertions, 35 (2%) reported 50 to 100 insertions, and 20 (1%) doctors reported over 100 insertions. The frequency and incidence of perforation events in each of these four groups is shown in Table 3.

Table 3. Frequency and incidence of uterine perforation according to number of insertions per doctor

	Insertions per doctor	Insertions performed	Perforation events	Perforation rate (per 1000 insertions)
	1–9	3634	11	3.0
	10–49	8297	11	1.3
	50–99	2402	1	0.4
	100 or more	2982	5	1.7

The perforation rate in the group where insertions were performed by doctors who reported less than 10 insertions was higher than the incidence of perforation in the group of doctors who reported performing 10 to 49 insertions (relative risk (RR) = 2.3, 95% confidence interval (CI) = 0.99–5.26, p = 0.046). It was also higher than the incidence in the 50–99 reported insertions group (RR = 7.3, 95% CI = 0.94–56.3, p = 0.03). Although the perforation rate in the group of doctors who reported performing less than 10 insertions was higher than for those who reported over 100 insertions, this difference was not statistically significant (RR = 1.8, 95% CI = 0.63–5.19, p = 0.26).

4. Discussion 

The incidence of uterine perforation in this study (when all possible cases were included) was 1.6 per 1,000 insertions, which is higher than the incidence reported in previous studies of Multiload Cu375 [1], [2], [3], [4]. This first raises the question whether it was correct to include all reports coded as ‘IUD embedded’ as uterine perforations. On examination of these case reports, some were suggestive of perforation at the time of insertion. Other reports, where the device was found to be embedded several years later, were more difficult to interpret. However, we considered that for an IUD to become embedded in the myometrium (at any stage after insertion), some degree of uterine wall damage is prerequisite. The total incidence of 1.6 per 1,000, therefore, includes partial as well as complete perforations.

There are other reasons why this study might have shown a higher perforation rate than others, including our first study of insertion events in this cohort of women, which reported a perforation rate of 0.6 per 1000 insertions [7]. In our previous study, fewer cases of perforation were identified as only events reported as occurring at insertion (or within 7 days) were searched for. In this subsequent study, many patients were followed for 4 to 5 years or longer and additional cases were identified. Follow-up in other studies has not been longer than 3 years [2], [3], which may not be long enough to identify some uterine perforations. In this study, the majority of perforations were not recognized at the time of insertion and some of the cases were not identified until years afterwards. In addition, larger studies should provide a more accurate estimate of the incidence of uncommon events. Earlier studies included fewer Multiload Cu375 insertions than this study—even when the results of several studies were pooled, the number of insertions was approximately 5,600 [4], which is less than one-third of the insertions included in our study.

The incidence of uterine perforation may also be higher in our study because the inserters were less experienced. The number of doctors performing insertions was far higher than in other studies [1], [2], [3] and the great majority (93%) of these doctors were not specialists. The results showed that 73% of all doctors reported inserting fewer than 10 Multiload Cu375 devices in the 10-year study period. The perforation rate in this group of insertions performed by less experienced doctors was significantly higher than the incidence of perforation reported by doctors who performed between 10 and 99 insertions. This result suggests that experienced inserters should fit IUDs. It is interesting, however, that the perforation rate in the group of the most experienced doctors (over 100 devices) was higher than the rate in the group of doctors whom reported performing between 50 and 99 insertions. This might be because of referral of difficult insertions to expert doctors in family planning clinics, for example.

Although the incidence of perforation is higher in our study than previous clinical trials, it is still possible that this is an underestimate of the true incidence of uterine perforation associated with IUD insertion. Some cases of perforation may have been reported as other events, for example, pain and bleeding or lost threads. The case reports from our study showed that some perforations were identified only after investigation of another problem. Where the need for this has not arisen, there may be other cases that remain undiagnosed. However, these factors may be balanced by a higher rate of reporting of uterine perforations in cases where the woman required hospital treatment.

Therefore, it is difficult to give an accurate estimate of the incidence of uterine perforation with Multiload Cu375 insertions. Our study suggests that this complication might be higher in ‘real-life’ clinical practice than in clinical trials. This might be expected as IUD insertions in New Zealand are generally performed outside specialized centers. Our results suggest that the perforation rate is higher when IUDs are fitted by doctors performing few insertions, which would indicate that experienced practitioners should perform insertions. However, an overall rate of less than two perforations per 1,000 insertions is still an acceptably low risk for most women and should not deter use of this effective method of contraception.