Prevalence and self-reported health consequences of vaginal practices in KwaZulu-Natal, South Africa: findings from a household survey

Authors

  • J. Smit,

    1.  Maternal, Adolescent and Child Health, Department of Obstetrics and Gynaecology, University of the Witwatersrand, Durban, South Africa
    2.  Reproductive Health and HIV Research Unit, Department of Obstetrics and Gynaecology, University of the Witwatersrand, Durban, South Africa
    Search for more papers by this author
  • M. F. Chersich,

    1.  Centre for Health Policy, School of Public Health, University of the Witwatersrand, Johannesburg, South Africa
    2.  International Centre for Reproductive Health, Department of Obstetrics and Gynaecology, Ghent University, Ghent, Belgium
    Search for more papers by this author
  • M. Beksinska,

    1.  Maternal, Adolescent and Child Health, Department of Obstetrics and Gynaecology, University of the Witwatersrand, Durban, South Africa
    2.  Reproductive Health and HIV Research Unit, Department of Obstetrics and Gynaecology, University of the Witwatersrand, Durban, South Africa
    Search for more papers by this author
  • B. Kunene,

    1.  Program for Appropriate Technology in Health (PATH), Johannesburg, South Africa
    Search for more papers by this author
  • N. Manzini,

    1.  Lovelife, Johannesburg, South Africa
    Search for more papers by this author
  • A. Martin Hilber,

    1.  Institute of Tropical and Preventive Medicine, Bern, Switzerland
    Search for more papers by this author
  • F. Scorgie

    1.  Maternal, Adolescent and Child Health, Department of Obstetrics and Gynaecology, University of the Witwatersrand, Durban, South Africa
    Search for more papers by this author

Corresponding Author Jennifer Smit, Maternal, Adolescent and Child Health (MatCH), Department of Obstetrics and Gynaecology, University of the Witwatersrand, 151 Juniper Road, Overport, Durban 4091, South Africa. Tel.: +27 31 275 1541; Fax +27 31 207 3239; E-mail: jsmit@match.org.za

Summary

Objectives  To investigate population-level prevalence of vaginal practices, their frequency and self-reported health consequences in KwaZulu-Natal, South Africa.

Methods  A household survey using multi-stage cluster sampling was conducted in 2007. Women aged 18–60 (n = 867) were interviewed on demographics, sexual behaviours and vaginal practices, focusing on intravaginal practices. Design-based analysis used multivariate logistic regression to identify factors associated with intravaginal or any practice.

Results  Most women currently perform vaginal practices (90.2%), with 34.8% reporting two and 16.3%≥3 practices. Internal cleansing, the commonest practice (63.3% of women), is undertaken frequently (61.6% cleansing twice daily; 20.0% using ≥2 products). Fewer report application (10.1%), insertion (11.6%) or ingestion (14.3%) practices. Hygiene is a common motivation, even for the 23.2% of women reporting intravaginal practices around the time of sex. Prevalence of any practice was lower among women with tertiary education than those without primary education (AOR = 0.26, 95% CI = 0.08–0.85), nearly twice as common in sexually active women (95% CI = 1.05–3.56) and increased as overall health status declined. Adjusted odds of intravaginal practices were 1.8-fold higher in women reporting unprotected sex (95% CI = 1.11–2.90). Few reported health problems with current practices (0.6%); though, 12.6% had ever-experienced adverse effects.

Conclusions  Vaginal practices are common in KwaZulu-Natal. Although self-reported health problems with current practices are rare, high lifetime risk of adverse events and potential for asymptomatic but clinically important damage make continued research important.

Abstract

Objectifs:  Examiner la prévalence à l’échelle de la population des pratiques vaginales, leur fréquence et leurs conséquences auto-rapportées sur la santé dans le KwaZulu-Natal, en Afrique du Sud.

Méthodes:  Une enquête sur les ménages utilisant l’échantillonnage en grappes à plusieurs niveaux a été réalisée en 2007. Les femmes âgées de 18 à 60 ans (n = 867) ont été interrogées sur les caractéristiques démographiques, les comportements sexuels et les pratiques vaginales, en se concentrant sur les pratiques intra vaginales. Une analyse fondée sur la régression logistique multivariée a été utilisée pour identifier les facteurs associés à la pratique intra vaginale ou à toute pratique.

Résultats:  La plupart des femmes effectuent couramment des pratiques vaginales (90,2%) avec 34,8% rapportant deux et 16,3% rapportant plus de trois pratiques. La douche intime, pratique la plus courante (63,3% des femmes), est effectuée fréquemment (61,6% y recourant deux fois par jour; 20,0% utilisant plus de 2 produits). Peu rapportent des pratiques par application (10,1%), insertion (11,6%) ou ingestion (14,3%). L’hygiène est une motivation commune, même pour les 23,2% de femmes déclarant des pratiques intra vaginales seulement autour des rapports sexuels. La prévalence de toute pratique était plus faible chez les femmes ayant une éducation tertiaire que chez celles sans éducation primaire (AOR = 0,26; IC95%: 0,08–0,85). Elle était près de deux fois plus élevée chez les femmes sexuellement actives (IC95%: 1,05–3,56) et augmentait lorsque l’état général de santé diminuait. Les rapports de cotes corrigés (OR) des pratiques intra vaginales étaient 1,8 fois plus élevés chez les femmes rapportant des rapports sexuels non protégés (IC95%: 1,11–2,90). Peu de problèmes de santé ont été rapportés avec les pratiques actuelles (0,6%), bien que 12,6% avaient déjà vécu des effets adverses.

Conclusions:  Les pratiques vaginales sont courantes dans le KwaZulu-Natal. Bien que les problèmes de santé auto-rapportés avec les pratiques actuelles sont rares, le risque élevé d’événements indésirables au cours de la vie et la possibilité de dégâts asymptomatiques mais cliniquement importants rendent la recherche utile.

Abstract

Objetivos:  Investigar la prevalencia poblacional de las prácticas vaginales, su frecuencia y consecuencias autorreportadas sobre la salud, en KwaZulu-Natal, Sudáfrica.

Métodos:  Estudio en hogares, utilizando un muestreo por conglomerados en varias etapas, durante el 2007. Se entrevistó a mujeres con edades comprendidas entre los 18–60 años (n = 867), sobre cuestiones demográficas, comportamiento sexual y prácticas vaginales, con un énfasis en las prácticas intravaginales. Se realizó un análisis mediante regresión logística múltiple para identificar los factores asociados con prácticas intravaginales o de otro tipo.

Resultados:  Actualmente la mayoría de las mujeres realizan algún tipo de práctica vaginal (90.2%), con un 34.8% reportando dos y 16.3%≥3 prácticas. La limpieza interna, la práctica más común (63.3% de las mujeres), es realizada frecuentemente (61.6% se limpiaban dos veces al día; 20.0% utilizando ≥2 productos). Un menor número reportó prácticas de aplicar (10.1%), insertar (11.6%) o ingerir (14.3%). La higiene es una motivación común, aún para el 23.2% de las mujeres que reportaban prácticas intravaginales cercanas al momento del sexo. La prevalencia de cualquier práctica era menor entre mujeres con una educación universitaria que entre aquellas con educación primaria (AOR = 0.26, 95%CI = 0.08–0.85), y casi dos veces más común entre mujeres sexualmente activas (IC 95% =1.05–3.56), y aumentaba a medida que disminuía el estatus general de salud. El riesgo relativo ajustado de las prácticas intravaginales era 1.8 veces mayor entre mujeres que reportaban la práctica sexo inseguro(95%IC = 1.11–2.90). Pocas mujeres reportaron problemas de salud relacionados con las prácticas vaginales (0.6%), aunque solo un 12.6% no habían experimentado nunca un efecto adverso.

Conclusiones:  Las prácticas vaginales eran comunes en KwaZulu-Natal. Aunque los problemas de salud autorreportados eran escasos, los riesgos de efectos adversos a largo plazo y el potencial de daños importantes, aunque asintomáticos, hace que la investigación continuada sea importante.

Introduction

Since the early 1990s, researchers in southern Africa have documented vaginal practices and suggested that some, particularly those for ‘dry sex’, are common (Smit et al. 2002; Runganga et al. 1992; Mbikusita-Lewanika et al. 2009; Braunstein & Van De Wijgert 2005; Beksinska et al. 1999; Brown & Brown 2000).Qualitative studies in this region have noted a diverse set of practices, in addition to ‘dry sex’, which include douching, application of substances to the external genitalia and various procedures for surgical or anatomical modification of women’s genitalia (Scorgie et al. 2009b; Bagnol & Mariano 2008). In South Africa, studies of high-risk groups or isolated communities found that the prevalence of vaginal practices varies from 26% in Cape Town (Myer et al. 2006) to 46% in an informal settlement near Johannesburg (Beksinska et al. 1999) to 80% in sex workers (Morar et al. 1998). The prevalence may be highest in KwaZulu-Natal Province (Baleta 1998).

Little is known about specific health-related effects of such practices. Early in the HIV epidemic, a few studies in Africa reported an association between some vaginal practices and HIV (Brown & Brown 2000; Mann et al. 1988). In other settings, researchers have also linked vaginal douching with sexually transmitted infections (STIs), bacterial vaginosis, candidiasis and infertility (Tsai et al. 2009; Joesoef et al. 1996; Baird et al. 1996; Martino & Vermund 2002). However, much uncertainty remains about which practices are harmful or might increase risk for HIV transmission, with longitudinal evidence scanty and findings inconsistent (Hilber et al. 2010). Moreover, harms are likely specific to a particular practice and to types of products and how they are used. More detailed information about patterns of use and motivations for vaginal practices would help to inform development of novel HIV prevention technologies and improve understanding of the uptake of existing ones.

While several household surveys in South Africa have documented sexual behaviours such as partner number, relationship type and condom use (Pettifor et al. 2005; Shisana et al. 2009), no research has yet described province-wide prevalence of vaginal practices and related aspects of sexual health in this setting. Absence of population-level data on vaginal practices (both those for sexual and for other purposes) is an important evidence gap as it is increasingly apparent that strategies to prevent HIV in generalised epidemic settings need to be informed by a comprehensive understanding of the sexual behaviour and health of the target group, tracked over time (WHO & UNAIDS 2002).This article reports on findings from a household survey in KwaZulu-Natal, South Africa, which specifically assessed vaginal practices in the context of women’s sexual well-being and health. The prevalence, intensity (frequency of practice and number of products used), motivations and self-reported health consequences of vaginal practices are presented here.

Methods

Study procedures

A household survey of women aged 18–60 years took place in the Province of KwaZulu-Natal, South Africa, between April and July 2007, using a multi-stage cluster sample design. Study activities were part of a multi-country WHO study, which also included one province each in Mozambique, Thailand and Indonesia (Hull et al. 2010). In South Africa, KwaZulu-Natal was selected on the basis of prior evidence of vaginal practices here and because the province has the highest HIV prevalence (Shisana et al. 2009; Department of Health 2009).

Development of the structured questionnaire was preceded by formative qualitative research, which ensured that language and expressions used in the generic questionnaire approximated those used in the study sites and that the options in categorical variables reflected local ranges, practices and meanings. These activities also assisted in deriving a classification of vaginal practices, detailed elsewhere (Hilber et al. 2007). A classification of products used by women for vaginal practices was also developed, consisting of ‘modern’ products, which are commercially produced and readily accessible from supermarkets, pharmacies or other retail outlets (e.g. detergents, talc, salt); ‘traditional’ products, which take the form of fruit, roots, bark and leaves, animal parts and ground minerals and are generally obtained from traditional healers, local retailers or informal traders1; and ‘commercialised traditional’ products, which are commercially manufactured, packaged and sold in a store, but marketed as ‘traditional’ in origin or content (e.g. bottled extract of African potato [Hypoxis hemerocallidea]) (Scorgie et al. 2009).

Survey tools were translated into isiZulu, back translated and pre-tested. To minimise potential non-response and measurement error, female interviewers were employed and trained to interview in a sensitive and non-judgmental manner in settings with adequate privacy. As available evidence about harms of specific vaginal practices is inconclusive, interviewers did not counsel women against such practices and avoided inferring that such practices were harmful. No remuneration was given for study participation. Respondents who reported conditions requiring medical attention were linked with local services.

Ethical approval was obtained from WHO and the University of the Witwatersrand. Prior to data collection, local community leaders were introduced to the study; they then facilitated entry to communities. Informed consent was first obtained from the household head, who identified eligible participants within the household and then additional consent obtained from the selected participant.

Sample design

The 2001 Census sampling frame was used (Statistics South Africa 2003), containing sampling units in the province (enumeration areas [EAs] and household listings). In a first stage of sampling, 28 EAs (primary sampling units) were selected with a probability proportional to EA size. Within selected EAs, 36 households were chosen using simple random sampling (1008 households in total). One woman per household was selected, with a Kish table used for sampling where households had more than one eligible woman (Kish 1949). If initial contact with the selected women failed, interviewers made a further two callbacks. Sampling occurred without replacement of either household or individual refusals. As prevalence of vaginal practices was estimated to occur in 50% of women (Beksinska et al. 1999), a sample of 850 completed interviews was chosen to give a precision of ±4% (estimated design effect = 2 and predicted non-response rate of 15%).

Study variables

Data were collected on population demographics, reproductive health and sexual behaviours, as well as on the broader social and gender context. Vaginal practices are defined here as those practices intended for physiological or anatomical modification of the labia, clitoris or vagina and are classified in seven main categories (Box S1) (Hilber et al. 2007). Women were questioned about each practice in turn, with these questions prefaced by a definition of the practice. Those who reported a current practice were classified as ‘current users’ in analysis. During data recoding, women were categorised as having performed external genital washing only if they reported doing the practice for reasons other than general bodily hygiene.

We examined the desired effects that women aim to achieve with vaginal practices, as well as their timing relative to daily hygiene, coitus, menstruation and other life events, such as childbirth. Self-reported health problems with current practice and history of adverse events were investigated, as well as variations in reported use of condoms and contraception with vaginal practices. ‘Transactional sex’ was defined as the receipt of gifts, payment or goods in exchange for sex in the past year.

Data management and analysis

Data were double entered by separate clerks at the local coordination centre, using Epidata version 3.1. Design-based analysis was performed using the survey components of STATA v8.2 (College Station, TX, USA). Given the unequal probabilities of selection, analysis corrected for the design effects of clustering of observational units and for sample weights (Korn & Graubard 1999). Person weights (reciprocal of selection probability) were used to adjust for differential probabilities of inclusion of EAs, number of eligible women per household and response rates. Summary indices for descriptive analysis are weighted point estimates and contingency table analysis used the Rao–Scott F statistic to determine P values (Rao & Scott 1984). Two multivariate logistic regression models were constructed to detect, first, factors associated with report of any vaginal practice and, second, factors associated with report of an intravaginal practice (intravaginal cleansing and/or insertion of products). Factors associated with dependent variables in univariate analysis (< 0.1) and age regardless of the presence of an association were entered into the model and retained if their removal markedly altered the model fit. Multivariate analysis used weighted maximum likelihood estimation with an adjusted Wald test F statistic.

Results

Study population

Nearly 90% of women eligible for the study were interviewed (867/972); 36 of the selected households had no eligible women. Median age of participants was 26.7 years, with about two-thirds living in urban areas (Table 1). For the majority of our sample, isiZulu was their home language (81.4%). Around 70% had some secondary school education, but only a further 9.7% had reached tertiary level.

Table 1.   Demographic and behavioural characteristics of women in household survey in KwaZulu-Natal, South Africa, weighted for household size and non-response
VariableAll women (%)
  1. Denominator is whole population unless indicated.

  2. *In women who had sex in past month.

  3. †Contraceptive use in women aged 18–49, sexually active and not currently pregnant.

  4. ‡Women who reported ever having signs of an abnormal discharge, multiple-response question.

  5. §Women who ever had sex.

Age
 18–2336.4
 24–2924.6
 30–4424.6
 45–6014.4
Area of residence
 Rural36.5
 Urban63.6
Language group
 Zulu81.4
 Xhosa6.0
 Other12.6
Level of education
 None/Primary incomplete13.8
 Primary complete4.7
 Some secondary71.8
 Tertiary9.7
Marital status
 Married18.3
 Unmarried, stable relationship60.4
 Other21.3
Sexually active (past month)62.9
Concurrent partnership18.4
Had transactional sex (past year)5.9
Condom use (past month)*
 None42.4
 Sometimes36.4
 Always21.3
Contraceptive use (past month)†
 None36.2
 Pill6.3
 Injectable23.1
 Male condom30.2
 Other3.9
Overall health status
 Good47.9
 Average42.5
 Poor9.6
Current genital illness or concern
 Any23.0
 Symptoms indicative of STI21.4
Signs of abnormal vaginal discharge (ever)47.1
Type of most recent discharge‡
 Excessive fluid12.3
 White discharge80.6
 Green discharge3.8
 Yellow discharge18.0
Ever had dyspareunia§22.7

A small portion was married (18.3%), although 60.4% were unmarried but in a relationship they described as stable. Fewer than 10% of women said they had never had sex; though, the majority reported between two and five lifetime sexual partners (55.7%). Around two-thirds of women were currently sexually active (62.9%). Almost 1 in 5 women who were married or in stable relationships reported currently having other partners (18.4%). Of women who had had sex in the past month, 21.3% said they had used condoms in all sex acts. Just under a third of sexually active women reported condom use at last sex (31.9%).

Contraceptive prevalence was 63.8%, with 23.1% reporting injectable contraceptive use and 30.2% consistent condom use in the past month. Three-quarters had previously been pregnant (74.7%), a median of two times (IQR = 1–3). Almost half reported ever having an abnormal discharge (47.1%), and 8.5% reported ever having genital ulcers or warts.

Prevalence and intensity of practices

Awareness of the full range of practices is generally high, although lowest for vaginal steaming or smoking (24.8%; Table 2), and anatomical modification (42.0%). For some practices (e.g. insertion), many more women had knowledge of the practice than had actually tried it. However, differences between being current and ever users of practices were smaller. Most women in KwaZulu-Natal currently undertake at least one vaginal practice (90.2%; Table 3). A third report two current practices (34.8%), and 16.3% three or more (data not shown).

Table 2.   Vaginal practices in KwaZulu-Natal, South Africa: prevalence, intensity of practices and reported effects weighted for household size and non-response
VariableExternal genital washing* (%)Internal Cleansing (%)Application (%)Insertion (%)Ingestion (%)Steaming or Smoking (%)Surgical or cutting %†
  1. *Excludes non-specific washing of the external genitalia that occurs as part of routine general bodily washing.

  2. †Lifetime use of surgical procedures. Denominator is whole population unless indicated.

  3. ‡In women reporting current practice.

  4. §Multiple-response question.

  5. ¶In women who ever used the vaginal practice.

Ever heard of practice74.954.467.865.124.842.0
Ever practiced72.164.412.016.721.37.63.0
Currently practice62.163.310.111.614.33.5
Frequency of current practice‡
 Median times in past month (IQR)60 (30–60)60 (60–60)12 (6–30)5 (4–10)8 (4–16)2 (1–5)1 (1–2)
Products used in past month‡
 146.780.094.496.791.989.392.4
 247.019.65.63.37.610.76.1
 ≥36.30.4000.501.6
Products most frequently used ‡,§Bath soap (77.6)
Household disinfectant (25.2)
Traditional (3.9)
Salt (2.2)
Cloth (52.3)
Paper (9.3)
Soap (4.8)
Household disinfectant(4.5)
Traditional (53.7)
Commercial creams (17.9)
Alum (9.0)
Talc (5.7)
Tampon (39.8)
Cloth (17.3)
Traditional (9.2)
Paper (6.2)
Talc (3.8)
Alum (3.1)
Traditional (66.5)
Not specified (18.9)
Commercial-traditional mixtures (6.7)
Salt (2.0)
Traditional (54.0)
Not specified (18.7)
Alum (8.0)
Commercial-traditional mixtures (8.9)
Vaginal cutting device (29.8)
Traditional (20.0)
Not specified (4.0)
Timing of practice§,¶
 During personal hygiene95.992.421.33.511.02.41.9
 Around menstruation32.616.98.335.96.032.20.0
 Preparation for sexual intercourse14.19.569.858.063.528.789.2
 After sexual intercourse14.218.94.95.08.41.41.2
 Time of physical discomfort5.04.25.75.415.28.60.0
 No pattern12.72.511.24.916.05.55.7
Motivation for practice‡,§
 Hygiene95.894.634.433.524.52.310.5
 Wellness83.138.232.310.419.220.00.0
 To treat infection or symptom20.326.029.126.029.485.40.0
 Sexual pleasure/partner commitment8.12.464.853.373.166.4100.0
 Pain during sex0.81.13.00.73.50.01.1
 Restore sensation of virginity2.91.51.62.76.30.04.6
 Feminine identity25.34.63.03.80.51.75.3
Self-reported effects‡,§
 Increased cleanliness97.389.841.832.626.69.45.7
 Odour reduction78.061.645.554.021.564.00.0
 Dry vagina14.448.414.922.735.254.17.0
 Lubricate vagina2.11.222.75.47.629.39.8
 Tighten vagina8.36.318.943.435.428.724.7
 Loosen vagina0.10.15.86.28.54.32.5
 Heat vagina1.20.621.230.13.53.739.2
 Stimulation0.90.04.21.02.54.90.0
 Symptom relief1.70.50.21.88.12.50.0
Health problem with current practice‡0.10.22.10.00.10.00.0
Adverse event with practice (ever)¶
 Any adverse event with practice7.44.711.719.11.79.94.2
 Genital irritation/pain2.93.65.57.00.07.33.3
 Genital itching4.20.75.85.00.11.51.0
 Bleeding or sores0.00.00.05.90.00.20.0
 Dyspareunia0.20.61.61.81.70.90.0
Table 3.   Multivariate logistic regression analysis of factors associated with report of any vaginal practices in KwaZulu-Natal, South Africa, weighted for household size and non-response
VariableAny current vaginal practice* (%)PUnadjusted odds ratio (95% CI)Adjusted odds ratio (95% CI)P
  1. The table presents prevalence of practices in each population strata (row percentages).

  2. *Excludes non-specific washing of the external genitalia that occurs as part of routine general bodily washing. Denominator is whole population unless indicated.

  3. †In women who had sex in past month.

  4. ‡Multiple-response question with a statistical test for each category, odds ratios compare with women not reporting that method.

  5. §Women who reported ever having an abnormal discharge, multiple-response question, with a statistical test for each category. Variables were included in the final multivariate models if associated with the dependent variable in univariate analysis (< 0.1) and if their removal markedly altered the model fit. Weighted maximum likelihood estimation was used with an adjusted Wald test F statistic.

Age
 18–2391.00.921.0
 24–2990.10.90 (0.37–2.19)
 30–4489.20.82 (0.44–1.52)
 45–6090.00.89 (0.39–2.08)
Area of residence
 Rural88.90.401.0
 Urban92.91.56 (0.54–4.52)
Language group
 Zulu91.20.161.0
 Xhosa92.41.17 (0.19–7.34)
 Other82.80.46 (0.24–0.90)
Level of education
 None/Primary incomplete93.7<0.0011.01.0 
 Primary complete97.93.12 (0.70–13.94)1.06 (0.52–2.16)0.86
 Secondary incomplete90.70.66 (0.30–1.45)0.75 (0.32–1.75)0.49
 Tertiary77.60.23 (0.08–0.66)0.28 (0.08–0.89)0.03
Marital status
 Married86.40.361.0
 Unmarried, stable relationship91.21.62 (0.94–2.79)
 Other90.71.54 (0.57–4.17)
Sexually active in past month91.70.051.72 (1.00–2.94)1.93 (1.05–3.56)0.04
Concurrent partnership88.80.580.83 (0.42–1.65)
Transactional sex in past year98.20.0096.67 (1.36–32.70)
Condom use in past month†
 Always88.30.471.0
 None or inconsistent92.60.48 (0.17–1.37)
Contraceptive use in past month‡
 None91.90.840.97 (0.48–1.96)0.03
 Pill82.80.240.51 (0.16–1.65)
 Injectable95.50.032.70 (1.14–6.38)2.54 (1.08–5.97)
 Male condom89.70.780.94 (0.47–1.89)
 Other93.50.581.58 (0.31–8.07)
Overall health status
 Good86.70.011.01.0 
 Average92.41.86 (0.99–3.49)1.56 (0.74–3.27)0.23
 Poor97.45.80 (1.42–23.68)3.92 (1.00–15.42)0.05
Current genital illness or concern
 Any96.00.053.07 (0.92–10.24)
 STI96.50.023.57 (1.13–11.31)2.68 (0.97–7.44)0.06
Abnormal vaginal discharge (ever)95.10.023.24 (1.17–8.97)  
 Type of recent discharge§
  Excessive fluid97.30.305.55 (2.34–13.17)
  White discharge97.2<0.0015.55 (2.34–13.17)
  Green discharge95.40.362.51 (0.30–20.83)
  Yellow discharge96.30.043.02 (0.99–9.28)
Ever had dyspareunia97.5<0.0015.61 (2.11–14.96)
Menstrual abnormality in past year
 None89.20.221.0
 Heavy bleeding90.71.18 (0.36–3.83)
 Lower abdominal pain91.71.34 (0.51–3.55)
 Amenorrhoea96.43.28 (1.26–8.52)

Internal cleansing, the commonest current practice, is very common: 61.6% cleanse twice daily, 15.5% more than twice and 20.0% report using two or more products (mainly cloth or paper; Table 2). Products chosen for this practice were predominantly used to achieve cleanliness (89.8%), odour reduction (61.6%) and vaginal drying (48.4%). External genital washing was practiced by 62.1% of women, which by definition (Box S1) (Hilber et al. 2007) is done for purposes other than hygiene, for example ‘wellness’ (83.1%), ‘feminine identity’ (25.3%) or to treat an infection or symptom (20.3%). Interestingly, 95.8% of women reported external genital washing said hygiene was a motive in addition to the other reasons listed. This practice is also common, with 59.5% cleansing twice daily, 12.9% more than twice a day and 53.3% report using two or more products, usually bath soap (77.6%) or household disinfectant (25.2%).

Intravaginal practices were currently undertaken by two-thirds of women, mostly internal cleansing alone (52.4%), with 10.9% performing both internal cleansing and insertion, and a few insertion alone (0.7%). Around one in ten currently practiced ingestion (14.3%) or application (10.1%), a median of 8 and 12 times a month, respectively (Table 2). In these practices, women generally used only one product, mostly traditional. Three women reported using tampons at times other than during menses to increase cleanliness, reduce odour and tighten the vagina. Women also insert traditional products (9.2%), cloth (17.3%), paper (6.2%), talc (3.8%) and alum (3.1%). Application, ingestion and insertion are frequently performed in preparation for sexual intercourse, with products often intended to tighten, dry or heat the vagina and to reduce odour (Table 2). A desire for additional lubrication also motivates some application (22.7% of current users). Few women have ever practiced vaginal steaming or smoking (7.6%) and even fewer surgical modification or cutting (3.0%). Overall, 23.2% of women reported an intravaginal practice around the time of sex, while 27.1% said that they used products to treat an infection or symptom.

Self-reported adverse effects

Few women reported health problems with current practices (0.6% with any current practice), although many more had ever-experienced side effects (12.6%). Adverse events were most common with insertion of substances, occurring in 19.1% of women who reported having ever done this (Table 2). These symptoms were predominately genital irritation (7.0%), vaginal bleeding or sores (5.9%) or itch (5.0%), while 1.8% linked this practice to dyspareunia. No side effects, however, were reported with current insertion. Similarly, although 11.7% had experienced side effects with application, few reported current effects (2.1%).

Association between vaginal practices and demographic, sexual and reproductive health factors

Prevalence of any current vaginal practice was similar among rural and urban women (Table 3). In multivariate analysis, compared with women who had no schooling or incomplete primary education (the baseline group), the odds of reporting any vaginal practice was 0.28 in women with tertiary education (AOR 95% CI = 0.08–0.89; = 0.03). As age was strongly associated with education level, we performed a separate analysis including education and age. This adjustment made negligible difference, however. Sexually active women were 1.9 times as likely to report any vaginal practice (AOR 95% CI = 1.05–3.56; = 0.04) as women who had had no sex in the past month. Almost all women who reported transactional sex currently have a vaginal practice (98.2%), just under half (46.7%) report two practices while 27.8% report three or more (data not shown). Women using injectable contraception had 2.5-fold higher odds of undertaking any current vaginal practice than other women (95% CI AOR = 1.08–5.97; = 0.03).

While no difference was detected in the overall prevalence of practices in different age groups, choice of practice varied. Just more than half of the women aged 30–44 reported intravaginal practices (55.7%), fewer than all other women (66.8%; = 0.007; Table 4). Women who said they used condoms inconsistently or not at all in the past month had 1.8 times higher odds of undertaking intravaginal practices than women reporting consistent condom use (AOR 95% CI = 1.11–2.90; = 0.02). In univariate analysis, compared with other types of vaginal discharges, women who described their most recent discharge as excessive or white in colour had higher prevalence of any current practice or intravaginal practices. Intravaginal practices were least common in women who reported having heavy menstrual bleeding, which concurs with the finding that vaginal practices generally are not undertaken around the time of menstruation (Table 2).

Table 4.   Multivariate logistic regression analysis of factors associated with report of an intravaginal practice in KwaZulu-Natal, South Africa, weighted for household size and non-response
VariableAny current intravaginal practice* (%)PUnadjusted odds ratio (95% CI)Adjusted odds ratio (95% CI)P
  1. The table presents prevalence of practices in each population strata (row percentages).

  2. *Internal cleansing and/or insertion practices. Denominator is whole population unless indicated.

  3. †In women who had sex in past month.

  4. ‡Multiple-response question with a statistical test for each category, odds ratios compare with women not reporting that method.

  5. §Women who reported ever having an abnormal discharge, multiple-response question, with a statistical test for each category. Variables were included in the final multivariate models if associated with the dependent variable in univariate analysis (< 0.1) and if their removal markedly altered the model fit. Weighted maximum likelihood estimation was used with an adjusted Wald test F statistic.

Age
 18–2363.30.031.0
 24–2970.01.35 (0.89–2.06)
 30–4455.70.73 (0.49–1.07)
 45–6069.91.35 (0.80–2.27)
Area of residence
 Rural63.90.971.0
 Urban64.11.00 (0.64–1.59)
Language group
 Zulu63.60.881.0
 Xhosa66.91.16 (0.59–2.26)
 Other65.61.09 (0.53–2.25)
Level of education
 None/Primary incomplete64.90.151.0
 Primary complete73.11.47 (0.66–3.25)
 Secondary incomplete65.11.00 (0.64–1.59)
 Tertiary50.20.55 (0.24–1.23)
Marital status
 Married65.80.751.0
 Unmarried, stable relationship64.40.94 (0.58–1.53)
 Other61.40.83 (0.52–1.33)
Sexually active in past month67.60.021.52 (1.09–2.12)1.35 (0.94–1.95)0.10
Concurrent partnership56.70.150.66 (0.36–1.18)
Transactional sex in past year84.60.033.18 (1.09–9.32)3.43 (1.24–9.52)0.02
Condom use in past month†
 Always55.10.061.01.00.02
 None or inconsistent70.91.96 (0.96–3.99)1.79 (1.11–2.90)
Contraceptive use in past month‡
 None62.70.691.04 (0.72–1.52)
 Pill51.60.090.59 (0.32–1.08)
 Injectable66.30.651.15 (0.64–2.04)
 Male condom66.00.721.09 (0.52–2.29)
 Other73.20.231.55 (0.71–3.37)
Overall health status
 Good60.10.051.0
 Average70.81.61 (1.08–2.42)
 Poor56.30.86 (0.40–1.81)
Current genital illness or concern
 Any80.0<0.0012.74 (1.82–4.13)2.94 (1.90–4.55)<0.001
 STI80.0<0.0012.69 (1.77–4.11)
Abnormal vaginal discharge (ever)77.8<0.0013.29 (1.88–5.75)  
 Type of recent discharge§
  Excessive fluid79.30.831.10 (0.45–2.71)
  White discharge81.20.022.49 (1.12–5.54)
  Green discharge37.70.020.16 (0.26–0.94)
  Yellow discharge61.90.030.37 (0.15–0.91)
Ever had dyspareunia74.80.011.90 (1.16–3.11)
Menstrual abnormality in past year
 None65.00.0031.0
 Heavy bleeding42.90.41 (0.24–0.68)
 Lower abdominal pain72.71.43 (0.75–2.72)
 Amenorrhoea70.51.29 (0.82–2.03)

Discussion and conclusions

Vaginal practices are reported by more than three-quarters of women in KwaZulu-Natal and are performed frequently and for very diverse purposes. Prevalence of practices was higher than previously reported in the general population in South Africa (Beksinska et al. 1999; Myer et al. 2006); though, directly comparing studies is difficult given marked variability of study measures. Higher prevalence among women who report transactional sex is consistent with previous evidence that practices are more common among sex workers than other population groups (Morar et al. 1998). Similar effects of educational level on likelihood of performing vaginal practices were noted in a previous study in South Africa (Beksinska et al. 1999). Interestingly, there is much commonality between the characteristics of those who have high prevalence of vaginal practices and those at raised risk for HIV. Identifying factors associated with higher prevalence of vaginal practices helps inform which women could be targeted with interventions to reduce vaginal practices, should future studies demonstrate harm more definitively (Chersich et al. 2009).

Adverse effects reported with current practices are rare, which is surprising and conflicts with findings of the qualitative study that preceded this survey (Scorgie et al. 2009b). That study captured detailed explanations of side effects associated with practices, both experienced and anticipated effects, although not always from women who actually reported experience of the practices. Contrasting findings are also potentially explained by the disparity between current and ever-experienced side effects, suggesting that if a woman has side effects, she may change to another product or dosage – or, indeed, cease the practice/s altogether. Importantly, the majority of adverse effects of vaginal practices may be asymptomatic epithelial disruption or inflammation, only visible on colposcopy or biopsy, as suggested by data in safety trials of potential microbicides. These studies have shown that compared to women receiving a placebo, genital lesions, mostly asymptomatic, were more common in women receiving nonoxynol-9, lime juice, PRO2000 and Savvy gel (Poynten et al. 2009). Given the high prevalence of vaginal practices found in our study, additional research and more robust methodologies are needed to more definitively determine harm of these practices and the products used, in particular, in regard to increased risk for HIV acquisition.

About 40% of women report a vaginal practice around the time of sex, showing the enormous potential for a microbicide product to interact chemically with commonly used vaginal products. Intravaginal cleansing, reported by around a quarter of women around the time of sex, in particular may result in sub-optimal concentration of some types of microbicides, which is most concerning as this effect may occur at the time of potential exposure to HIV during sex. Where disinfectant is used for such cleansing (instead of soap, for example), additional harm could potentially be done. Equally concerning is the finding that condom use was lower among women who report vaginal practices, implying that any resulting genital inflammation or lesions will thus be occurring in women more likely to have unprotected sex. Previous qualitative research in the same population has also indicated that practices to alter the vaginal environment, making it tight, dry and hot, are at odds with condom use (Scorgie et al. 2009b). The MIRA phase III diaphragm trial conducted in South Africa and Zimbabwe similarly found that vaginal practices were associated with a lower likelihood of women consistently using condoms and the diaphragm (Van Der Straten et al. 2010).

The association between vaginal practices and injectable contraception, which has been reported previously (Smit et al. 2002), suggests that uptake of contraceptive and HIV prevention technologies, both present and future ones, are likely affected by local notions of vaginal health. A more nuanced understanding of women’s preferred vaginal state to feel clean, healthy, sexually appealing and prepared for sexual relations is important for informing the optimal consistency and formulation of contraceptives and potential microbicide products (Braunstein & Van De Wijgert 2005). Indeed, that women are familiar with intravaginal products – and, in many cases, use them to improve sexual pleasure, health and overall wellbeing– suggests that there might be opportunities to build on this familiarity when promoting microbicides (Montgomery et al. 2010).

Although complex methodologically and logistically, use of population-level data is a strength of this study. The population demographics are similar to those of previous household surveys in KwaZulu-Natal, suggesting sampling frames were appropriate (Department of Health & Measure DHS 2007). There are limitations to the survey design, however, making inferences about causal relationships invalid. For example, vaginal practices were more common in women with previous vaginal discharge, but the temporal relationship between practices and discharge cannot be determined from this study (Myer et al. 2006). Additionally, adverse events were all self-reported and not assessed as ‘current’ events, thus the study design could not adequately assess the safety of the various practices reported.

This study – the first to measure prevalence of vaginal practices at a population level in KwaZulu-Natal – shows that while these practices, with a wide range of products, are common in this province, much still remains unknown about their safety. Without more definitive information on how harmful these practices and products are, however, such queries are likely to remain unresolved. Further research on this topic is all the more urgent, given recent findings on the efficacy of an antiretroviral-based microbicide candidate in the Caprisa 004 trial (Abdool Karim et al. 2010).The design of effective interventions to improve women’s sexual health and well-being, much needed in this region, should take into account both the adverse effects of vaginal practices, where these exist, and build on the benefits as perceived by the women who undertake them. Further, note should also be taken of our finding, supported by other studies, that women reporting unprotected sex were more likely to report intravaginal practices. This too should be addressed in the development of effective interventions aimed at preventing HIV.

Footnotes

  • 1

    The actual chemical composition of these products is not known and users know these products by name only, not by what they are made of.

Acknowledgements

The Multi-country study on gender, sexuality and vaginal practices (GSVP) was a project of the UNDP/UNFPA/WHO/World Bank Special Programme on Research, Development, and Research Training in Human Reproduction. Generous support has been provided by AusAID, the Australian Research Council, the Flemish Government, the Ford Foundation, the International Partnership on Microbicides, UNAIDS and WHO. The authors acknowledge the input of the South African study participants and the study team. We thank Helen Rees for her comments on the article.

Ancillary