Abstract
Introduction: To investigate the predictors factors of Pelvic Organ Prolapse (POP) in Iranian women by using ordinal logistic regression.
Materials and Methods: The role of risk factors of POP was evaluated among 365 patients attending in two public centres in Ilam province, Iran. Exclusion criteria were of single, pregnant and lactate women receiving hormone replacement therapy. Both the unvaried and multi-variate ordinal logistic regression were used to find the predictive factors of POP and computing sensitivity and specificity of models.
Results: In multi-variate ordinal logistic regression the variables of Body Mass Index (BMI) Maximum Birth Weight (MBW) and delivery mode were the most important factors for prediction of prolapse stage. The sensitivity and specificity of multi-variate ordinal logistic, as a screening test, were 95.7% and 48.7% respectively.
Conclusion: BMI, MBW and delivery mode can use for prediction of POP stage. Stage of prolapse is an ordinal variable, therefore to show relationship between stage of prolapse and other variables ordinal logistic regression is an appropriate model.
Keywords: Ordinal Logistic Regression, Pelvic Organ Prolapse, Dorsal Lithotomy position ligament
Introduction
Pelvic Organ Prolapse (POP) is defined as the descent of intra-pelvic organs due to deficiencies in the pelvic support system. It is a common condition among women [1,2]. POP can affect the women of all ages and is associated with functional problems of the pelvic floor [3].
The worldwide prevalence of genital prolapse is estimated to be 2–20% in women under the age of 45 years [4]. However, it is common among elderly woman and the frequency has been reported as high as 39.8% [5]. The lifetime risk of women undergoing surgery for incontinence or prolapse is estimated to be 11% in the USA [6]. Based on previous studies conducted in Iran, the prevalence of POP is higher than the global situation, so that about half of the Iranian women, who experience some degree of POP [7].
POP is an important public health issue and it has side effects on quality of life and impact on the health care system [8]. On the other hand, the results of the studies indicate that 135, 000 women undergo surgery for UI and 225,000 have POP repair each year in the United States [9,10].
There are many techniques of classifying POP. The Baden-Walker (grades 0 through 4) and Pelvic Organ Prolapse–quantification POPQ (stages 0 through IV) are the two main systems for staging the degree of POP [4]. However, no unique system is generally agreed upon, the system approved by the International Continence Society (ICS), called the POPQ system, is considered as one standard. This POPQ system contributed significantly to the development in studying prolapse because it allows researchers to report findings by a standardized style [11].
There are common risk factors for POP, including aging, pregnancy, vaginal birth, birth trauma, chronic increases intra-abdominal pressure (obesity, chronic constipation, chronic coughing, repetitive heavy lifting), menopause, estrogen deficiency, genetic factor, prior surgery, myopathy, collagen abnormalities, smoking and lifestyle [3,12,13]. In prior, studies have used of X2, t-test, liner and logistic regression to find relationship between prolapse stage and risk factors. Since the stage of prolapse is an ordinal variable, the aim of present study is investigated the predictive factors for POP by using of Ordinal Logistic Regression in Iranian women’s.
Materials and Methods
This is a cross-sectional study carried out to examine the role of demographic, anthropometric and clinical variables on POP. In the first, we evaluated 1,000 available Health Information documents in two primary health care centers in Ilam province, Iran. In the second, extracted address and phone number of eligible women for the study. Then invited the eligible women to participate in the study according to a schedule. All the women participated in the study except for single, pregnant and lactate women receiving hormone replacement therapy. Overall, in this research 365 women participated.
Data collection and examinations was carried out in face to face interviews by researcher.
Demographic information: This included data on age, education, and occupation.
Anthropometrics data: This included information on weight and height. Trained research personnel measured height and weight by a Seca 220 (made by Germany) while the subjects were minimally clothed and not wearing shoes. The BMI was calculated based on heights and weights [BMI = weight (kg)/(height (m)2]. Based on the BMI, women were grouped into different categories as recommended by the National Centre for Education in Maternal and Child Health.
Clinical measures: The summary of the patient’s obstetrical included data on pregnancy, delivery mode, delivery operative, delivery position, MBW, medical and surgical histories and POP type and stage collected with observation, interview, and examination.
All the subjects evaluated in Dorsal lithotomy position after emptying their bladders according to the POPQ technique recommended by ICS. In this system measurements are made at different vaginal sites, providing quantification of prolapse affecting different vaginal segments (anterior and posterior vagina, vaginal apex or cervix) as well as an overall stage of prolapse. Measurements are made in centimeters relative to the hymen as the reference point. In this system, negative numbers represent positions above the hymen, and positive numbers represent points beyond or past the hymen. A rigid marked hysterometer, calibrated in centimeters was used for measurements. Also, the small vaginal dilator was used to measure Aa, Ba, Ap‚ Bp. The apical points of C, D and the Total Vaginal Length (TVL) points were measured relative the hymen. All points except for TVL were recorded in maximal valsalva effort. The external measurements of GH and PB made at the time of rest and with strain. Then, bimanual examination for determine abdominal and pelvic mass were performed.
Patients were divided into two groups for comparison: (1) with prolapse (2) without prolapse. Both univariate and multiple logistic regression analyses were used to indicate the association between dependent (with prolapse vs. without prolapse) and independent variables. Independent variables tested for an association were age, education, occupation, BMI, pregnancy, delivery mode, delivery operative, delivery position, and MBW. Significant level was considered to p=0.05.
The Ethics Committee of Ilam University of Medical Sciences approved the study design. Written informed consent was obtained from the participants after comprehensive explanation of the procedure involved.
Results
Of the 365 women examined in this study, 285(80.8%) women were put in the POP group. The overall distribution of POP-Q system stages were as follows: stage 0, 19.2%, stage I, 20% (73/365) and stage II, 60.8% (222/365) .The prevalence of POP per stage in relation with independent variables in our general population is presented in [Table/Fig-1].
[Table/Fig-1]:
Severity stages population according to selected independent variables
| Severity stage 0 | Severity stage 1 | Severity stage 2 | p-value | |
|---|---|---|---|---|
| Age (Mean ± SD) | 32.19 ±7.16 | 34.88 ±9.22 | 38.02 ±8.99 | 0.000 |
| Body mass index (Mean±SD) | 26.29 ± 4.78 | 27.67 ± 4.84 | 29.02 ± 4.91 | 0.000 |
| Maximum birth weight (Mean ±SD) | 3443.75 ±548.09 | 3340 ±781.56 | 3787 ± 663.25 | 0.000 |
| Pregnancy number(Mean±SD) | 3.17 ± 2.78 | 3.81 ± 2.99 | 5.37± 2.99 | 0.000 |
| Constipation N(%) | 0.103 | |||
| Yes | 5 (8.9) | 12 (21.4) | 39 (69.6) | |
| No | 65 (21) | 61 (19.7) | 183 (59.2) | |
| Pelvic Surgical N(%) | 0.005 | |||
| Yes | 19 (28.4) | 19 (28.4) | 29 (43.3) | |
| No | 51 (17.1) | 54 (18.1) | 193 (64.8) | |
| Occupation N(%) | 0.002 | |||
| Light work | 58 (24.1) | 50 (20.7) | 133 (55.2) | |
| Heavy work | 12 (9.7) | 23 (18.5) | 89 (71.8) | |
| Delivery operative N(%) | 0.000 | |||
| Mix | 16 (28.6) | 8 (14.3) | 32 (57.1) | |
| Traditional Midwife | 7 (6.4) | 17 (15.6) | 85 (78) | |
| Physician | 24 (54.5) | 12 (27.3) | 8 (18.2) | |
| Obestetrician | 16 (12) | 26 (19.5) | 91 (68.4) | |
| None | 7 (30.4) | 10 (43.5) | 6 (26.1) | |
| Delivery position N(%) | 0.000 | |||
| Mix | 5 (9.8) | 9 (17.6) | 37 (72.5) | |
| Home | 7 (6.4) | 17 (15.6) | 85 (78) | |
| Hospital | 51 (28) | 37 (20.3) | 94 (51.6) | |
| None | 7 (30.4) | 10 (43.5) | 6 (26.1) | |
| Delivery Mode N(%) | 0.000 | |||
| Normal Vaginal Delivery and other | 39 (13) | 53 (17.7) | 208 (69.3) | |
| CS + Nulli gravid | 30 (46.9) | 20 (31.2) | 14 (21.9) |
The present study participants had a relatively higher rate of anterior vaginal prolapse (sub stage AA, Be: 72.3%) than other sub stages (Ape, Bp: 49.9%, C: 47.1% and all points 28.77%). There is significant association between increasing age with increasing stage of POP (38.72 ± 8.94 in prolapse group vs 32.76 ± 7.29 in the without prolapse group, p= 0.000). Of all participants 82 % (300/365) had experienced at least one vaginal delivery. Association prolapse stages with other factors using univariate ordinal logistic regression analysis are presented in [Table/Fig-2].
[Table/Fig-2]:
Association prolapse stage with other factors using univariate ordinal logistic regression analysis (link function complementary log-log)
| B | SE | Wald | p-value | |
|---|---|---|---|---|
| Age | 0.05 | 0.01 | 22.65 | 0.000 |
| Body mass index | 0.08 | 0.02 | 17.11 | 0.000 |
| Maximum birth weight | 0.001 | 0.00 | 10.48 | 0.001 |
| Pregnancy number | 0.182 | 0.031 | 33.5 | 0.000 |
| Constipation | ||||
| Yes | 0.408 | 0.26 | 2.43 | 0.12 |
| No | 1.0 (ref.) | |||
| Pelvic Surgical | ||||
| Yes | -0.64 | 0.19 | 11.35 | 0.001 |
| No | 1.0 (ref.) | |||
| Occupation | ||||
| Light work | -0.61 | 0.19 | 9.86 | 0.002 |
| Heavy work | 1.0 (ref.) | |||
| Delivery operative | ||||
| Mix | 0.73 | 0.32 | 5.11 | 0.02 |
| Traditional Midwife | 1.61 | 0.33 | 24.56 | 0.000 |
| Physician | -0.39 | 0.30 | 1.66 | 0.2 |
| Obestetrician | 1.19 | 0.3 | 16.03 | 0.000 |
| None | 1.0 (ref.) | |||
| Delivery position | ||||
| Mix | 1.34 | 0.37 | 13.2 | 0.000 |
| Home | 1.60 | 0.33 | 24.05 | 0.000 |
| Hospital | 0.58 | 0.27 | 4.43 | 0.03 |
| None | 1.0 (ref.) | |||
| Delivery Mode | ||||
| NVD and other | 1.44 | 0.18 | 64.3 | 0.000 |
| CS + Nulli gravid | 1.0 (ref.) |
The coefficient of ordinal logistic regression for age shows that the increasing of age is associated with increasing stage of Pelvic Organ Prolapse. The wald statistic in [Table/Fig-2] show that the importance of prolapse of variable in prediction os stage of prolapse. Fore example, variables of pregnancy number and delivery mode are more important than others variables in predication of prolapse stage.
In multi-variate ordinal logistic regression the variables of BMI, MBW and delivery mode were the most important variables for prediction of prolapse stage. The wald statistic in [Table/Fig-3] show that delivery mode, BMI, and MBW are the most important variables in prediction of prolapse stage respectively.
[Table/Fig-3]:
Association prolapse stage with the covariates using multi-variate ordinal logistic regression analysis (link function complementary log-log)
| Estimate | Std. Error | Wald | p | |
|---|---|---|---|---|
| Threshold | ||||
| Stage0 | 3.856 | 1.072 | 12.937 | 0.000 |
| Stage1 | 4.652 | 1.086 | 18.360 | 0.000 |
| Location | ||||
| Body mass index | 0.093 | 0.031 | 8.810 | 0.003 |
| Maximum birth weight | 0.000 | 0.000 | 4.006 | 0.045 |
| Delivery mode: NVD | 1.518 | 0.268 | 32.119 | 0.000 |
| Delivery mode: CS | 1.0 (ref.) |
If we consider multi-variate ordinal logistic, as a screening test, the sensitivity of test is high (95.7%) but the specificity of test is low (48.7%) [Table/Fig-4].
[Table/Fig-4]:
Sensivity and spesifity of the ordinal logistic model
| Predicted Response | ||||
|---|---|---|---|---|
| Stage 0 | Stage 2 | Total | p | |
| Obseved Response | Stage 0 | 19 (48.7) | 20 (51.3) | 39 (100) |
| Stage 1 | 9 (36) | 16 (64) | 25 (100) | |
| Stage 2 | 4 (4.3) | 88 (95.7) | 92 (100) | |
| Total | 32 (20.5) | 124 (79.5) | 156 (100) | |
Discussion
In this cross-sectional study, according to the POPQ system the majority (80.8%) of our study population met criteria for pelvic organ prolapse. Several studies have shown a wide range of POP ranging from 2-2.6% in the general population [13], 4-12.2% in Sweden [14], 11.4% amoung women aged 45-85 years [15], 19.7% among women in 16 low-income and lower middle-income countries [16] and 39.8% in women who enrolled in the Women’s Health Initiative Hormone Replacement Therapy (HRT) Clinical Trial [5]. The prevalence of POP was higher in our study (80.8%) as compared to other study conducted in Iran (53%) [7]. Regarding the effect of age, lifestyle, number and type of delivery and other risk factors [17,18], differences between the research participants can be attributed to the difference in the reported prevalence.
We found that stage 2 is most common severity of prolaps. In Other Iranian study most participants have stage 1 or 2 of POP [7]. Trowbridge et al., confirm our results and reported the stage 2 in 67.7% of all participants [19].
In present study, we investigated a model for predication of POP. The univariate ordinal logistic regression showed significant statistical association between age, BMI, MBW, pregnancy number, pelvic surgical, occupation, delivery operative, and delivery mode with stage of POP. There was any association between constipation and stage of pelvic organ prolapse. In a case-control study evaluated the risk factors for the development of genital prolapse in the Brazilian population. The age, BMI, parturition, number of vaginal, caesarean section or forceps deliveries, newborn weight and positive family history factors in patients were factors associated with POP. After Logistic Regression presence of at least one vaginal delivery, fetal macrosomia and positive family history for dystocia have reported as independent risk factors in POP [20]. Although several studies have confirmed the effect of increasing age on the incidence and severity of POP [9,21–24], in a case study reported the third-degree uterine prolapse in 11-month-old infant [25].
Based on best knowledge, it is the first study that use ordinal logistic regression to analysis data. In previous research to find association between prolapse stage and other variables used X2 and t-test and liner regression and logistic regression [7,26–27]. Stage of prolapse is an ordinal variable so that to show relationship between it and other variables the appropriate model is ordinal logistic regression. In Ordinal Logistic Regression finding appropriate link function is very important.
Conclusion
In present research, we choose much link function such as Cauchit, Complementary log-log, Logit, Negative log-log and Probit. Likelihood function shows that complementary log-log is the best link function for this data. In Ordinal Logistic Regression we can compute specificity and sensitivity of model that can have medical application and help to midwife and Gynaecologist to diagnosis of stage of prolapse.
Financial or Other Competing Interests
None.
References
- [1].Voorham-van der Zalm PJ, Lycklama ANGA, Elzevier HW, Putter H, Pelger RC. Diagnostic investigation of the pelvic floor: a helpful tool in the approach in patients with complaints of micturition, defecation, and/or sexual dysfunction. J Sex Med. 2008;5(4):864–71. doi: 10.1111/j.1743-6109.2007.00725.x. [DOI] [PubMed] [Google Scholar]
- [2].Patel DA, Xu X, Thomason AD, Ransom SB, Ivy JS, DeLancey JO. Childbirth and pelvic floor dysfunction: an epidemiologic approach to the assessment of prevention opportunities at delivery. Am J Obstet Gynaecol. 2006;195(1):23–8. doi: 10.1016/j.ajog.2006.01.042. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Low LK, Tumbarello JA. Falling out: Authoritative knowledge and women’s experiences with pelvic organ prolapse. J Midwifery Wom Heal. 2012;57(5): 489–94. doi: 10.1111/j.1542-2011.2012.00187.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Kuncharapu I, Majeroni BA, Johnson DW. Pelvic organ prolapse. Am Fam Physician. 2010; May 1;81(9):1111–7. [PubMed] [Google Scholar]
- [5].Schon Ybarra MA, Gutman RE, Rini D, Handa VL. Etiology of post-uterosacral suspension neuropathies. Int Urogynaecol J Pelvic Floor Dysfunct. 2009;20(9):1067–71. doi: 10.1007/s00192-009-0894-x. [DOI] [PubMed] [Google Scholar]
- [6].Olsen AL, Smith VJ, Bergstrom JO, Colling JC, Clark AL. Epidemiology of surgically managed pelvic organ prolapse and urinary incontinence. Obstet Gynaecol. 1997;89(4):501–6. doi: 10.1016/S0029-7844(97)00058-6. [DOI] [PubMed] [Google Scholar]
- [7].Garshasbi A, Faghih-Zadeh S, Falah N. The status of pelvic supporting organs in a population of iranian women 18 - 68 years of age and possible related factors. Arch Iran Med. 2006;9(2):124–8. [PubMed] [Google Scholar]
- [8].Bump RC, Norton PA. Epidemiology and natural history of pelvic floor dysfunction. Obstet Gynaecol Clin North Am. 1998 Dec;25(4):723–46. doi: 10.1016/s0889-8545(05)70039-5. [DOI] [PubMed] [Google Scholar]
- [9].Waetjen LE, Subak LL, Shen H, Lin F, Wang TH, Vittinghoff E, Brown JS. Stress urinary incontinence surgery in the United States. Obstet Gynaecol. 2003;101(4):671–76. doi: 10.1016/s0029-7844(02)03124-1. [DOI] [PubMed] [Google Scholar]
- [10].Hamilton-Boyles S, Weber AM, Meyn L. Prolapse procedures in the United States, 1979–1997. Am J Obstet Gynaecol. 2003;188:108–15. doi: 10.1067/mob.2003.101. [DOI] [PubMed] [Google Scholar]
- [11].Weber AM, Abrams P, Brubaker L, Cundiff G, Davis G, Dmochowski RR, et al. The standardization of terminology for researchers in female pelvic floor disorders. Int Urogynaecol J Pelvic Floor Dysfunct. 2001;12(3):178–86. doi: 10.1007/PL00004033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Ygaard I, Bradley C, Brandt D. Pelvic organ prolapse in older women: prevalence and risk factors. Obstet Gynaecol. 2004;104(3):489–97. doi: 10.1097/01.AOG.0000136100.10818.d8. [DOI] [PubMed] [Google Scholar]
- [13].Samuelsson EC, Victor FT, Tibblin G, Svardsudd KF. Signs of genital prolapse in a Swedish population of women 20 to 59 years of age and possible related factors. Am J Obstet Gynaecol. 1999;180(2 Pt 1):299–305. doi: 10.1016/s0002-9378(99)70203-6. [DOI] [PubMed] [Google Scholar]
- [14].Tegerstedt G, Maehle-Schmidt M, Nyren O, Hammarstrom M. Prevalence of symptomatic pelvic organ prolapse in a Swedish population. Int Urogynaecol J Pelvic Floor Dysfunct. 2005;16(6):497–503. doi: 10.1007/s00192-005-1326-1. [DOI] [PubMed] [Google Scholar]
- [15].Slieker-ten Hove MC, Pool-Goudzwaard AL, Eijkemans MJ, Steegers-Theunissen RP, Burger CW, Vierhout ME. Symptomatic pelvic organ prolapse and possible risk factors in a general population. Am J Obstet Gynaecol. 2009;200(2):184e1–7. doi: 10.1016/j.ajog.2008.08.070. [DOI] [PubMed] [Google Scholar]
- [16].Walker GJ, Gunasekera P. Pelvic organ prolapse and incontinence in developing countries: review of prevalence and risk factors. Int Urogynaecol J. 2011;22(2):127–35. doi: 10.1007/s00192-010-1215-0. [DOI] [PubMed] [Google Scholar]
- [17].Quiroz LH, Chang H, Blomquist JL, Okoh YK, Handa VL. Scheduled cesarean delivery: maternal and neonatal risks in primiparous women in a community hospital setting. Am J Perinatol. 2009;26(4):271–7. doi: 10.1055/s-0028-1103155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Nygaard I, Bradley C, Brandt D. Pelvic organ prolapse in older women: prevalence and risk factors. Obstet Gynaecol. 2004;104(3):489–97. doi: 10.1097/01.AOG.0000136100.10818.d8. [DOI] [PubMed] [Google Scholar]
- [19].Trowbridge ER, Fenner DE. Conservative management of pelvic organ prolapse. Clin Obstet Gynaecol. 2005;48(3):668–81. doi: 10.1097/01.grf.0000171740.27555.f1. [DOI] [PubMed] [Google Scholar]
- [20].Odrigues AM, de Oliveira LM, Martins Kde F, Del Roy CA, Sartori MG, et al. Risk factors for genital prolapse in a Brazilian population. Rev Bras Ginecol Obstet. 2009;31(1):17–21. doi: 10.1590/s0100-72032009000100004. [DOI] [PubMed] [Google Scholar]
- [21].Kim CM, Jeon MJ, Chung DJ, Kim SK, Kim JW, Bai SW. Risk factors for pelvic organ prolapse. Int J Gynaecol Obstet. 2007;98(3):248–51. doi: 10.1016/j.ijgo.2007.02.019. [DOI] [PubMed] [Google Scholar]
- [22].Rodrigues AM, de Oliveira LM, Martins Kde F, Del Roy CA, Sartori MG, et al. Risk factors for genital prolapse in a Brazilian population. Rev Bras Ginecol Obstet. 2009;31(1):17–21. doi: 10.1590/s0100-72032009000100004. [DOI] [PubMed] [Google Scholar]
- [23].Dietz HP. The aetiology of prolapse. Int Urogynaecol J Pelvic Floor Dysfunct. 2008;19(10):1323–9. doi: 10.1007/s00192-008-0695-7. [DOI] [PubMed] [Google Scholar]
- [24].Kepenekci I, Keskinkilic B, Akinsu F, Cakir P, Elhan AH, Erkek AB, et al. Prevalence of pelvic floor disorders in the female population and the impact of age, mode of delivery, and parity. Dis Colon Rectum. 2011;54(1):85–94. doi: 10.1007/DCR.0b013e3181fd2356. [DOI] [PubMed] [Google Scholar]
- [25].Chien WH, Tseng MR, Lin YH. Acute pelvic organ prolapse in an 11-month-old infant. Int UroGynaecol J Pelvic Floor Dysfunct. 2006;17(5):548–9. doi: 10.1007/s00192-005-0026-1. [DOI] [PubMed] [Google Scholar]
- [26].Handa VL, Nygaard I, Kenton K, Cundiff GW, Ghetti C, Ye W, et al. Pelvic organ support among primiparous women in the first year after childbirth. Int UroGynaecol J Pelvic Floor Dysfunct. 2009;20(12):1407–11. doi: 10.1007/s00192-009-0937-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Wang J, Varma MG, Creasman JM, Subak LL, Brown JS, Thom DH, et al. Pelvic floor disorders and quality of life in women with self-reported irritable bowel syndrome. Aliment Pharmacol Ther. 2008;31(3):424–31. doi: 10.1111/j.1365-2036.2009.04180.x. [DOI] [PMC free article] [PubMed] [Google Scholar]