Factors Associated with Loss to Follow-up Among Acceptors of the Levonorgestrel-releasing Intrauterine System at A Public-sector Family Planning Clinic in Brazil
*Corresponding author:Dr. Anibal Faúndes, Department of Obstetrics and Gynecology, School of Medical Sciences, State University of Campinas, Caixa Postal 6181, 13083-970 Campinas, SP, Brazil,E-mail: firstname.lastname@example.org
In view of its confirmed effectiveness and long-lasting action, the 52-mg levonorgestrel-releasing intrauterine system (LNG-IUS) [1-2] is recognized as one of the most effective long-acting reversible contraceptives (LARCs). The LNG-IUS has also been found to be effective for the treatment of heavy menstrual bleeding (HMB). In addition, the device has been successfully used for treating endometriosis-associated pain, dysmenorrhea, endometrial hyperplasia and cancer, although it remains off-label for these indications [3-5].
A limitation to the broader use of the LNG-IUS concerns its upfront cost; however, the manufacturer, together with the Population Council, created the International Contraceptive Access (ICA) Foundation, which provides large quantities of the device to non-profit institutions such as the Campinas Center for Reproductive Health Research (CEMICAMP). This initiative allows this method to be provided at no cost at the family planning clinic of the Department of Obstetrics and Gynecology, State University of Campinas and at another 28 centers located throughout Brazil.
The LNG-IUS has been very well accepted by the clients of this family planning clinic, both as a contraceptive alone and as a treatment for HMB and dysmenorrhea. However, a large proportion of LNG-IUS acceptors, over 25%, never return to the clinic for follow-up.
Women lost to follow-up (LFU) are usually treated in analysis as if they were no different from those who continue under observation. It is assumed, for example, that accidental pregnancy rates in these women and in those who remain under observation are similar. Nevertheless, although the device may be less likely to fail in women who become LFU, the contrasting hypothesis is also possible. Although this potential source of bias in the evaluation of contraceptives has long been recognized , insufficient attention has been given to this problem and little is known about the factors associated with loss to follow-up.
Since the method is expensive but offered free of charge at this clinic, it is possible that some women attend the clinic for the sole purpose of obtaining the method and later continue their care in another public service institute, health maintenance organization (HMO) or private healthcare service, hence failing to return to this clinic for follow-up. To obtain further information regarding this problem, the present study was designed to determine the factors associated with loss to follow-up in LNG-IUS acceptors at UNICAMP’s family planning clinic.
Material and methods
This was a descriptive, comparative, retrospective cohort study conducted at the Department of Obstetrics and Gynecology, University of Campinas School of Medical Sciences, Campinas, São Paulo, Brazil. The institute’s internal review board approved the study protocol. No data were collected that would allow any of the subjects to be identified, thus ensuring the confidentiality of this information. Since this was a retrospective study based on data obtained from patients’ medical records and since the identity of each woman was maintained confidential, the requirement for informed consent was waived.
All the women who initiated use of the 52-mg LNG-IUS between January 2007 and April 2015 were included in the study. It proved impossible to perform a calculation of sample size, since the investigators were unable to identify any other studies on the subject that could serve as a basis on which to estimate expected differences. However, as 5,916 women were followed up for 15 months following device insertion, it is fairly certain that differences would not have to be large to be identified in this study.
The main outcome measure of the study was the rate of LFU, and the independent variables were age at device insertion, parity, marital status (dichotomized as married/in a stable union or single/divorced/separated/data unavailable); years of schooling, and why the LNG-IUS was indicated (categorized as for contraception alone, for medical reasons alone or for both).
The investigators obtained data directly from the pre-coded clinical records at the family planning clinic and from two special forms required by the ICA Foundation that donated the LNG-IUS: one to be completed at admission and the other at each follow-up visit. Two independent data entry clerks input the data in duplicate, and any inconsistencies were reviewed and corrected.
Acceptors of the device were requested to return for a follow-up visit at one year after insertion; therefore, women who had not returned for follow-up by fifteen months after insertion were defined as LFU in accordance with the generally accepted recommendations .
The rates of loss to follow-up according to the various age groups, parity, marital status, years of schooling and the reason for indicating the LNG-IUS for each woman were compared to determine whether there was an association between these factors and the main outcome.
Logistic regression analysis was performed to evaluate the relationship between the dependent dichotomous variable, LFU, and the independent variables listed above . Bivariate and multivariate analyses were conducted. Odds ratios (OR) and their 95% confidence intervals (95%CI) are reported for the predictors that were statistically significant. The entire statistical analysis was performed using the SAS software package, version 9.4 (SAS Institute, Cary, North Carolina, USA).
Characteristics of the LNG-IUS acceptors in the sample
A total of 5,916 women opted to use an LNG-IUS during the study period. Their sociodemographic characteristics are described in Table 1,
as well as the reason for the indication of the LNG-IUS. Almost two-thirds of the sample was 30 years of age or more, and indeed one-third of the patients were at least 35 years of age. Almost 40% had between 9 and 12 years of schooling, while 32% had 8 years of schooling or less, and 20% had more than 12 years of formal education.
Over three-quarters of the sample requested the LNG-IUS exclusively for contraception, while of the remaining women a similar number were using it for medical reasons and for both contraceptive and medical reasons.
Loss to follow-up as a function of the women’s sociodemographic characteristics and the reason for indicating the LNG-IUS
At 15 months after insertion, the rate of LFU was highest among women under 25 years of age and lowest among those over 35 years of age. The difference was statistically significant only when the lowest age group, which consisted of women ≤24 years of age, is compared with the highest age group, consisting of women ≥35 years of age. No association was found between parity or marital status and the rate of LFU.
There was an increase in the LFU rate as a function of the women’s education level, with higher LFU rates being associated with a greater number of years of schooling. The odds ratio for being LFU was twice as high among the women who had at least 12 years of schooling compared to those with 8 years of schooling or less.
The rate of LFU was significantly higher among women using the LNG-IUS exclusively for contraception compared to those using it for medical reasons or for both contraceptive and medical reasons.
All the associations found in the bivariate analysis persisted after the multiple regression statistical analysis, as shown in Table 3.
The present results tend to confirm the hypothesis that women with more years of schooling, a good marker of socioeconomic level, are more likely to become LFU compared to those with fewer years of schooling in this population of LNG-IUS acceptors. Our original hypothesis was that these women are often clients of an HMO or patients of physicians with a private office, and possibly request the LNG-IUS at the university’s family planning clinic due to the high cost of obtaining the device privately. These women would, however, return to their original physicians for follow-up rather than continuing to be monitored at the family planning clinic. If that hypothesis is correct, the same association will not be found in other settings where cost is not a factor that has to be taken into consideration when selecting the method, such as in the population of a US-based study in which participants were provided with reversible contraception of their choice at no cost .
To the best of our knowledge, only one study has evaluated associations between the education level of LNG-IUS/copper IUD users and loss to follow-up . Those authors found that women who complied with follow-up visits were similar insofar as education was concerned to those who became lost to follow-up. Regrettably, no other publications were found in which education was investigated as a factor involved in LFU, with the remaining studies evaluating the effect of education level on discontinuation of use rather than LFU. If discontinuation were to be taken as an acceptable indicator of LFU, then the present results showing a higher rate of LFU in women with better education would contradict previous findings showing that the discontinuation rate was lower among more educated women or that there was no association between discontinuation and education level [10,11].
This difference between the results of the present study and the findings reported from other studies [10,11] may be solely due to the use of discontinuation as an indicator of LFU, which ultimately may not be correct. However, it may also be because, in the specific case of this clinic, women with higher socioeconomic levels, who are not part of our usual clientele, come to the clinic for the sole purpose of obtaining the LNG-IUS free of charge and are not interested in being followed up at the university clinic.
The finding that the LFU rate was higher among younger women compared to older users is in agreement with the few publications dealing with this issue. In those papers, younger women were associated with higher rates of LFU during the use of contraceptives, including IUDs [12,13].
A finding that is common to the majority of studies is that younger age is associated with a greater likelihood of the woman discontinuing use of the method, with the age-related difference increasing as a function of the period of observation after insertion [14,15].
There was no association between parity and LFU. This finding agrees with the results of another study  but contradicts an older study carried out in Mexico, which found that lower parity was associated with higher rates of LFU . On the other hand, several studies have found discontinuation rates to be higher in women with lower parity, including nulligravidas [10,16,17]. However, the fact that around 80% of the women in the present study had had only one or two children may explain why no association was found between parity and LFU.
Our finding that the LFU rate was almost twice as high in women using the method exclusively for contraception compared to those using the device for the treatment of gynecological problems was unexpected. We were unable to find any publications analyzing this factor as a reason for discontinuation and can only speculate that many women come to the university clinic to obtain an LNG-IUS, but then go on to seek other public clinics closer to their homes for their follow-up care. Women using the LNG-IUS for medical reasons may feel that it is safer to continue attending the university clinic rather than their neighborhood primary care clinic, where the same level of specialized care would not be available.
The scarcity of publications dealing with the problem of loss to follow-up is noteworthy. Perhaps, the rates are too low in high-income countries to cause concern. In a large study carried out in Denmark, Finland and Sweden, for example, the rate of LFU was about 11% after three years . In other settings, however, an LFU rate of over 36% after one year of use was reported in a study conducted in Peru  and a rate of 38.6% was reported by a study carried out in Pennsylvania, USA with women who initiated use of the LNG-IUS after a second trimester abortion .
Trussel  mentions another study in which Tietze found a 49% LFU rate. After performing a search to find those women who were LFU, the pregnancy rate for the contraceptive under evaluation increased from 9.4 to 14.4/100 women-years, showing how misleading the results of a study may be when too many users are LFU .
Only one publication was found in which means of reducing LFU rates were evaluated . Those investigators found that providing an e-mail address, having a prepaid or contract cell phone, having text-messaging capabilities and a preferred means of contact other than the telephone were factors associated with successful contact at the 12-month follow-up visit .
The 26% LFU rate found for the present study population is far from encouraging and stimulates us to initiate a similar study in this clinic to test different means of contacting patients who miss their scheduled visits. Until now, the method used here has been the telephone, which, according to Torres et al. , is the least effective means of re-establishing contact, indicating a need to identify other means of reducing LFU in this setting.
Although the importance of keeping the LFU rate low in clinical studies of any contraceptive is clear, one may question the purpose of asking users of LNG-IUS/copper IUDs to return to the clinic for follow-up when they are not volunteers in any clinical trial. Recently, Dragoman et al.  challenged the scientific community with two questions: “Does checking for the presence of an IUD string(s) improve user satisfaction with the method?” and “Does checking for the presence of an IUD string(s) increase the method’s effectiveness by alerting the user when/if an expulsion has occurred?”. To the best of our knowledge, there are no answers to those questions yet. Many years ago, Rivera et al.  proposed that for women who are not volunteers in a clinical trial, there is no need for further follow-up after IUD insertion except if complications occur, since recommendations to return for periodic follow-up visits only increase the number of unnecessary consultations. In the case of the family planning clinic at this department, where the performance of the principal contraceptive methods is a matter of continuous evaluation, every effort must be made to keep the LFU rate low.
The main limitation of the present study concerns the fact that the LNG-IUS is provided at no cost at this clinic, which is not the case in other settings in Brazil, in the rest of the region or indeed in the world. Therefore, recommendations based on the results of the present study cannot be extrapolated to clinics in which conditions differ from those found here.
The finding that women with higher education levels are at a significantly greater risk of loss to follow-up should serve as a warning to the many other clinics that receive this device from the ICA Foundation. The need to counsel women in that group on the importance of returning to the clinic for follow-up consultations should be emphasized. We also expect that calling attention to this particular problem will contribute towards increasing the importance given to identifying ways of reducing the proportion of acceptors that are LFU in any family planning clinic interested in evaluating their results.
Conflict of interest
LB is an advisory member of the Board of Trustees of the ICA Foundation. The other authors have no conflicts of interest to declare.
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
- Luukkainen T, Allonen H, Haukkamaa M, et al. Effective contraception with the levonorgestrel-releasing intrauterine device: 12-month report of a European multicenter study. Contraception 1987; 36(2):169-179.
- Faúndes A, Alvarez F, Díaz J. A Latin American experience with levonorgestrel IUD. Ann Med 1993;25(2):149-153.
- Luukkainen T, Toivonen J. Levonorgestrel-releasing IUD as a method of contraception with therapeutic properties. Contraception 1995;52(5):269-276.
- Bahamondes L, Ribeiro-Huguet P, de Andrade KC, et al. Levonorgestrel-releasing intrauterine system (Mirena) as a therapy for endometrial hyperplasia and carcinoma. Acta Obstet Gynecol Scand 2003;82(6):580-582.
- Bahamondes L, Petta CA, Fernandes A, et al. Use of the levonorgestrel-releasing intrauterine system in women with endometriosis, chronic pelvic pain and dysmenorrhea. Contraception 2007;75(6 Suppl):S134-9.
- Trussell J. Methodological pitfalls in the analysis of contraceptive failure. Stat Med 1991;10(2):201-220.
- Tietze C, Lewit S. Recommended procedures for the statistical evaluation of intrauterine contraception. Stud Fam Plann 1973;4(2):35-42.
- Agresti A. Categorical data analysis. New York: John Wiley & Sons; 1990.
- Winner B, Peipert JF, Zhao Q, et al. Effectiveness of long-acting reversible contraception. N Engl J Med 2012;366:1998-2007.
- Phillips SJ, Hofler LG, Modest AM, et al. Continuation of copper and levonorgestrel intrauterine devices: a retrospective cohort study. Am J Obstet Gynecol 2017; 217(1):57.e1-57.e6.
- Diedrich JT, Zhao Q, Madden T, et al. Three-year continuation of reversible contraception. Am J Obstet Gynecol 2015;213(5):662.e1–8.
- Keller A. Mexico City: A clinic dropout study. Stud Fam Plann 1971;2:192-196.
- Torres LN, Turok DK, Sanders JN, et al. We should really keep in touch: predictors of the ability to maintain contact with contraception clinical trial participants over 12 months. Contraception 2014;90(6):575-580.
- Backman T, Huhtala S, Blom T, et al. Length of use and symptoms associated with premature removal of the levonorgestrel intrauterine system: a nation-wide study of 17,360 users. BJOG 2000;107(3):335-339.
- Escudero F, Gonzales GF, Delgadillo L, et al. Factors associated with discontinuation rates of the Copper T380A in a Peruvian public hospital. Adv Contracept 1999;15(4):303-311.
- Abraham M, Zhao Q, Peipert JF. Young age, nulliparity, and continuation of long-acting reversible contraceptive methods. Obstet Gynecol 2015;126(4):823–829.
- Bateson D, Harvey C, Trinh L, et al. User characteristics, experiences and continuation rates of copper intrauterine device use in a cohort of Australian women. Aust N Z J Obstet Gynaecol 2016;56(6):655–661.
- Nygren KG, Nielsen NC, Pyörälä T, et al. Intrauterine contraception with Nova-T and copper-T-200 during three years. Contraception 1981;24(5):529-542.
- Hohmann HL, Reeves MF, Chen BA, et al. Immediate versus delayed insertion of the levonorgestrel-releasing intrauterine device following dilation and evacuation: a randomized controlled trial. Contraception 2012;85(3):240–245.
- Tietze C, Poliakoff SR, Rock J. The clinical effectiveness of the rhythm method of contraception. Fertil Steril 1951;2(5):444-450.
- Dragoman MV, Jatlaoui T, Nanda K, et al. Research gaps identified during the 2014 update of the WHO medical eligibility criteria for contraceptive use and selected practice recommendations for contraceptive use. Contraception 2016;94(3):195-201.
- Rivera R, Chen-Mok M, McMullen S. Analysis of client characteristics that may affect early discontinuation of the TCu-380A IUD. Contraception 1999;60(3):155-160.