Sex Transm Inf 1998;74:35–39 35 Clinical algorithms for the screening of pregnant Sex Transm Infect: first published as 10.1136/sti.74.1.35 on 1 February 1998. Downloaded from Original women for STDs in Libreville, Gabon: which article alternatives? A Bourgeois, D Henzel, G Malonga-Mouelet, G Dibanga, C Tsobou, M Peeters, E Delaporte Objective: Sexually transmitted diseases (STDs) remain one of the major public health problems in the developing world. To implement a systematic screening of STDs among pregnant women in Libreville, Gabon, a preliminary cross sectional study on STD prevalence and risk factors was performed in antenatal clinics. A score, integrating risk factors and elementary clinical signs for the screening of STDs, showed higher performances compared with hierarchical algorithms. The prospective validation of this score based on six criteria (risk factors and simple clinical signs) was done in 1994–5. The sensitivity (76.7%), compared with results from other studies, was accept- able for diagnosing cervical infection (Neisseria gonorrhoeae and/or Chlamydia trachomatis) but the specificity was low (50.6%). In addition, the diagnostic values for diagnosing vaginal infection (Trichomonas vaginalis and/or Candida albicans) were poor. We then proposed to evaluate an alternative flowchart for the screening of cervical and vaginal infections. Methods: In this study, 646 pregnant women were enrolled. Each woman was interviewed and examined by a physician and then was subjected to reference laboratory examinations. An algo- rithm in two steps, combining a risk assessment score at the beginning of a hierarchical process, followed by a second step more specifically applied to a limited number of women, was developed and evaluated. Results: The prevalence rate was 11.3% for cervical infection and 39.5% for vaginal infection. The first step of the algorithm, applied to all pregnant women, is based on four criteria (age, marital status, dyspareunia, coloured vaginal discharge). It allows classification of the women into three classes: high, low, and intermediate risk of cervical infection. Only the patients with inter- mediate risk were submitted to further investigations including speculum and microscopic examination, and subsequently chlamydial antigen detection. This flowchart was 83.6% and 81.2% sensitive and 63.4% and 62.7% specific for predicting cervical infection and vaginal infec- tion, respectively. Conclusion: Similar strategies using simple rapid tests for chlamydial and gonococcal infection would certainly constitute a good diagnostic tool. This theoretical model needs to be evaluated prospectively, not only to confirm their diagnostic value but also to evaluate their feasibility, reli- http://sti.bmj.com/ ability and acceptability, as well as their cost eVectiveness. (Sex Transm Inf 1998;74:35–39) Keywords: algorithm; Africa; antenatal clinic; STDs on September 29, 2021 by guest. Protected copyright. Programme SIDA de Introduction syndromic approach is that patients with l’ORSTOM, Sexually transmitted diseases (STDs) remain asymptomatic infections are not detected. Montpellier, France one of the major public health problems in the Furthermore, diVerent reports on the evalu- A Bourgeois developing world. This is so because STDs ation of algorithms for vaginal discharge have M Peeters enhance the sexual transmission of HIV,12and shown a poor diagnostic value even using a risk E Delaporte also because of the morbidity and mortality score, so the use of such algorithms remains rates of STDs among women and children.3–6 questionable.11 12 INSERM U13, Paris, 78 France Sociocultural and clinical factors pose In Gabon, STD prevalence rates among preg- D Henzel particular challenges for the control of STDs. In nant women are very high and the infertility rate developing countries, moreover, the lack of due to STDs is one of the highest observed in Programme National laboratory facilities or qualified staV and limited Africa.13–15 This justifies the systematic screening de Lutte contre le financial resources are other constraints. for the most important and frequent STDs, SIDA, Libreville, In such situations, the World Health which include gonococcal, chlamydial, tri- Gabon 16 G Malonga-Mouelet Organisation (WHO) has recommended for chomonal, and candidal infections. G Dibanga the management of STDs the syndromic Even if Candida albicans infections are C Tsobou approach,910 which allows the treatment of vaginal infections more often and not sexually patients at the first consultation. For this transmitted, they are classically considered Correspondence to: syndromic approach, an algorithm is used with other STDs, in particular in the WHO E Delaporte, Programme 910 SIDA, 911 Avenue which can be hierarchical, based on successive management guide for STDs. As one objec- Agropolis, BP 5045, 34032 and logical steps, or non-hierarchical based on tive of this study was the evaluation of the Montpellier cdx 1, France. a risk factors score. This approach is not WHO recommendation we kept to this Accepted for publication dependent on the laboratory results, and could classification. In order to implement systematic 29 August 1997 be cost eYcient. The major drawback of the STD screening among pregnant women visit- 36 Bourgeois, Henzel, Malonga-Mouelet, et al ing dispensaries in Libreville, the capital of for examination and (or) treatment. The Gabon, a preliminary cross sectional study was women could also receive the treatment for performed. From this study, a flowchart based their partners. Sex Transm Infect: first published as 10.1136/sti.74.1.35 on 1 February 1998. Downloaded from on a score favouring the diagnosis of cervical infection was selected. The prospective evalua- PRINCIPLE USED FOR THE CONCEPTION OF NEW 17 tion of this flowchart was done in 1994–5 ALGORITHMS among pregnant women presenting for their The previous evaluated flowcharts16 17 had lim- first antenatal visit. ited specificity for diagnosing cervical infection In this population, gonococcal or chlamydial and a low sensitivity for the diagnosis of vagin- cervical infection was found in 11.3% of the itis. The main objective of such flowcharts is to cases, and trichomonal or candidal vaginal detect cervical infections with a high sensitivity infection in 39.5% of the cases. The perform- because of their complications. However, the ances of this score were unsatisfactory for diag- detection of vaginal infections (being very nosing cervical infection with a sensitivity of prevalent) is also important. To increase the 76.7% and a specificity of 50.6%. Positive and specificity of the flowchart for diagnosing the negative predictive values were 17.3% and cervicitis, without reducing the sensitivity, we 94.1% respectively. developed algorithms in two steps, combining a Based on this result, and before the wide- risk assessment score at the beginning of a spread use of this strategy, we proposed to hierarchical process, followed by a second step evaluate a new algorithm with higher specificity more specifically applied to a limited number for cervical infection and higher sensitivity for of women. vaginal infection. We report here the develop- The first step, applied to all pregnant ment of this new algorithm in two steps, com- women, aims to classify the patients into three bining a risk assessment score and a hierarchi- categories: patients with low, intermediate, or cal process. high probability of cervical infection. This step is based on a risk assessment score using only Methods very simple criteria as risk factors or simple POPULATION clinical signs chosen for their feasibility and The study was conducted among pregnant reproducibility by all midwives. Women with women attending three antenatal clinics in high risk of cervicitis will be treated immedi- Libreville between November 1994 and March ately for cervical infection, whereas women 1995. During this period, all women presenting with low risk will be further investigated for consecutively for their first antenatal visit were vaginitis only. Women with intermediate risk recruited. Well informed consent was obtained will be submitted to a second step, for the before inclusion in this study. In all the clinics search of additional signs of cervical infection, each woman was interviewed and examined by as well as vaginal infection. The criteria used in the physician. From the interview, demo- this second step have to be more specific for the graphic data, sexual and gynaeco-obstetric his- diagnosis of cervicitis, and more sophisticated tory, and current STD symptoms were col- tests, depending on the availability of facilities, http://sti.bmj.com/ lected. Physical examination included can be used. inspection with a speculum. Serum samples and cervicovaginal swabs were collected for the DATA ANALYSIS reference laboratory tests. DBASE and EPI-INFO were used for entry, management, and data analysis. DiVerences 2 DEFINITIONS AND LABORATORY DIAGNOSIS in rates were tested with ÷ test. Intensity of link Cervical infection was restricted to gonococcal between characteristics and cervical or vaginal on September 29, 2021 by guest. Protected copyright. and/or chlamydial infection, vaginitis to candi- diseases was estimated by the odds ratio. dal and/or trichomonal infection.910 N gonor- rhoeae was identified by culture on modified Results Thayer–Martin medium or by presence of GENERAL CHARACTERISTICS Gram negative diplococci at direct examin- In this study, 646 pregnant women were exam- ation of a cervical smear. A second endocervi- ined by