A study on usefulness of a set of known risk factors in predicting maternal syphilis infections in three districts of Western Province, Zambia

A study on usefulness of a set of known risk factors in predicting maternal syphilis infections in three districts of Western Province, Zambia

Jacob Sakala^1,2, Nellisiwe Chizuni¹, Selestine Nzala¹

 

¹Department of Public Health, University of Zambia, School of Medicine, Lusaka, Zambia, ²Kaoma District Medical Office, Ministry of Health, Kaoma, Zambia

 

 

^&Corresponding author
Jacob Sakala, Department of Public Health, University of Zambia, School of Medicine, Lusaka, Zambia

 

 

 

 

An estimated 11 million people are infected with syphilis globally every year; of these 1.5 million are pregnant women [1]. Maternal syphilis can lead to adverse pregnancy outcomes (APO) in 53.4 - 81.8% cases [2] such as spontaneous abortions, intra-uterine growth retardation, still births, premature deliveries, low-birth weight, perinatal deaths and congenital disease among new-born babies [3, 4]. To contribute towards control of maternal syphilis and its APOs, the World Health Organization (WHO) recommends antenatal screening of all pregnant women [5]. Many countries in sub-Saharan Africa (SSA) including Zambia, which have some of the highest levels of infection, introduced the WHO syphilis-screening recommendations into national antenatal care (ANC) guidelines [6, 7]. However, most of these SSA countries have challenges ensuring 90% antenatal access to syphilis screening services. In 2010, of the 27 countries in SSA that submitted a report, only Namibia achieved this target. The median coverage for syphilis screening in the reporting countries was 59% [8]. Zambia, on average, screens 44% pregnant women every year and the rural areas of the western province are among the worst affected. As a result of this, an estimated 241 prenatal syphilis infections are missed annually in three districts (Kaoma, Luampa and Nkeyema) of this province alone [9, 10]. To ensure universal access to antenatal syphilis testing, a number of studies have recommended the use of point-of-care (POC) biomedical tests [1,11, 12]. To this effect, in 2012, Zambia introduced more cost effective and logistically simpler rapid syphilis tests (RST) [3, 6] to the existing antenatal screening guidelines that originally provided only for use of rapid plasmin reagin tests (RPR) [13]. Despite this, there has been minimal improvement in screening levels indicating existence of possible performance gaps related to health systems, political support and resources allocation for the programme [14,15]. Understandably, to address existing performance gaps, the main focus for research in this area has been developing more cost-effective, biomedical means of screening for syphilis as infections can be asymptomatic. We argue that a need exists to find alternative methods for use in identifying women at high risk of infection when biomedical test are unavailable. Currently, ANC providers receive no guidance from the WHO and national guidelines on alternative measures to control maternal syphilis when biomedical tests are unavailable for any reason. A few studies, whilst in the process of conducting cost-effectiveness studies of biomedical testing, explored alternative control-strategies such as empirical treatment of all pregnant women [16-18] and presumptive syphilis diagnosis based on presence of genital ulcer disease (GUD) [11]. Results have shown that these alternative methods could be relatively cheaper than conventional methods, however, concerns of unnecessary over-treatment of antenatal attendees have discouraged their recommendation for general use [18]. Arising from this, we sought to explore the possibility of using a clinical protocol, based known risk factors of maternal syphilis, as an alternative syphilis identification method. The rationale was based on availability of documented evidence of risk factors associated with maternal syphilis which could be categorized in a clinical protocol. These risk factors included; a maternal history of previous infection with syphilis, history of abortion [19], history of multiple sexual partners [20], early maternal age at sexual debut [21], obstetric history of still birth delivery [22, 23], HIV co-infection [24], presence of genital ulcer disease [11, 25] and a history of sexually transmitted infection in the partner [26]. We then evaluated the sensitivity, specificity and predictive values of the designed protocol. We further, sought to review how these findings would influence performance gaps in antenatal syphilis screening guideline implementation in three districts of western province, Zambia.

 

 

Design and sampling procedures

 

We conducted a cross-sectional study in Kaoma, Luampa and Nkeyema districts, western province, Zambia from April to June 2015. This is the most populated region of the province with 122,092 inhabitants, majority of whom are women of child bearing age with annual expected pregnancies of 10,531 and average first antenatal attendances of 9,989. The region has 34 public health facilities offering ANC services. The primary sample was selected using one-stage cluster design which guided the initial selection of eight (8) health facility clusters using systematic random technique from a list public health facilities in the study region. Respondents were then recruited by selecting all consenting pregnant women attending antenatal clinic in the sampled clusters during the study period. The secondary sample included 10 health facility staff from these sites as well as 3 district health programme managers selected purposefully. We collected data on past exposure to risk factors of maternal syphilis from primary respondents. The RPR blood test performed on consenting respondents was gold standard syphilis test. We also performed a record review of; antenatal records, antenatal screening policy guidelines, district health systems in relation staff capacities and logistics management and latter included interviews of district health staff.

 

Data collection

 

A structured interviewer-administered questionnaire was used for data collection. The tool was translated into Lozi, the main local language in the study region, and pre-tested for consistency. The regular ANC providers were recruited as research assistants and the questionnaire was administered in normal antenatal clinic setting to gain respondents' trust especially that some questions could be considered sensitive.

 

Laboratory methods

 

Blood screening for syphilis was done after administration of the questionnaire to minimise bias. The study provided test kits to facilities when unavailable. The test was conducted under the usual antenatal conditions using the IMMUTREP RPR, a non-treponemal flocculation syphilis test. Approximately 50µl venous whole blood sample from each consenting respondent was mixed with one free-falling drop of test antigen on a test card. The mixing was aided by rotating the test cards for 8 minutes after which the results were read.

 

Proposed clinical assessment criteria

 

Risk factors were categorized into major factors which included: a maternal history of previous infection with syphilis, presence of genital ulcer disease, history of multiple sexual partners, HIV co-infection and minor factors; history of abortion, early maternal age at sexual debut, obstetric history of still birth delivery and neonatal death, as well as a history of sexually transmitted infection in the partner. This classification was based on observed strength of association or frequency of linkage of these risk factors to maternal syphilis infections with cut-off set at odds ratio of 5 and frequency of 10%. Clinical disease was then defined as either the presence of two (2) major risk factors or one (1) major and two (2) minor risk factors or three (3) minor risk factors.

 

Data analysis

 

The data from the questionnaires was coded, checked and cleaned before entry into a Microsoft excel sheet and imported into Stata version 13 for analysis. Data on risk factor identification and laboratory results were assigned numbers either 1 or 2 depending on presence or absence of a risk factor or disease. These were then entered in Microsoft excel and imported into Stata version 13. Proportions were used to estimate prevalence of maternal syphilis. Since data variables of risk factors was dichotomous, univariate and multivariate analysis for binary outcomes was done to find the relationship to maternal syphilis with odds ratio and chi-square as measures of association. Multiple logistic regression was done to test for significance set at 95% confidence level with p value < 0.05. Using RPR as confirmatory test, the sensitivity, specificity and predictive values of the proposed criteria were calculated to measure its usefulness in identifying maternal infections. The accuracy of the assessment criteria was ascertained by calculating the area under the receiver operating curve (ROC) which compared ability of a test to differentiate between those with disease and those without. A ROC of greater than 0.80 was deemed to have good accuracy, while 0.70 to 0.80 was fair and less than 0.70 was deemed to be poor. We used content analysis method to summarize qualitative data from desk review and health personnel unstructured interviews. This was then reported by use of narratives which in some cases included direct key quotations.

 

Ethical Considerations

 

Ethics approval was obtained from Excellence in Research Ethics and Science (ERES) Converge; (2014-Aug-016). Participation was consensual and data was handled confidentially.

 

 

Population characteristics

 

A total 740 respondents who gave complete responses were recruited in the study. The majority; 360 (48.6%) were aged between 20 and 30 years with a mean age of 26 ± 0.5 years. The results also showed that a greater proportion of the respondents; 445 (60.1%) were married, had more than one pregnancy; 572 (77.3%) and were unlikely to have gone beyond primary level of school education; 497 (67.2%).

 

Syphilis sero-positivity in study population

 

The syphilis sero-positivity using RPR tests was observed to be 9.3% (95% CI: 7.4 -11.6%) among study participants. Of the women testing positive, 61 (87.1%) were multigravidas as opposed to 9 (12.9%) who were in their first pregnancy.

 

Correlates between risk-factors and maternal syphilis infections among respondents

 

A univariate comparison of risk factors to maternal syphilis showed that cases were more likely to have a history of abortion, a history of still birth delivery, to have previously lost a baby in the first month of birth, to have previous infection with syphilis, to have a sexual partner with a sexually transmitted infection, to have had multiple sexual partners in past (two) 2 years, to have genital ulcer disease and be HIV co-infected. However, the association between maternal syphilis infection and early sexual debut (before the age of 16 years) at this stage was found not to be statistically significant as the chi-square test for this association was found to have a P-value greater than 0.05 (Table 1). We then conducted a multivariate logistics analysis excluding early age at sexual debut. After controlling for all variables we found that co-morbid conditions of HIV and genital ulcer disease and exposure histories of still birth delivery and previous infection with syphilis were strongly associated with syphilis sero-positivity (OR>5). Other risk factors such as history of abortion, having more than one sexual partners and sexually transmitted infection in a sexual partner were also significantly associated with gestational syphilis infections (OR:3 to 5). At this stage, we also found that a history of losing a neonate through death was not significantly associated to maternal syphilis (OR 2.3, p value> 0.05) (Table 2). Therefore only seven (7) of the nine (9) preselected factors in the end were found to be significantly associated with maternal syphilis.

 

Sensitivity, specificity and predictive value of proposed risk assessment criteria

 

The proposed assessment criteria identified 59 (8%) of the respondents with presumptive clinical disease. Of these, 43 were true positive (TP) cases of syphilis. The criteria also identified 655 women as true negatives (TN). However, 26 (37.7%) women with disease were missed and 27.1% were incorrectly classified as diseased when they were syphilis sero-negative. The overall sensitivity of the assessment criteria was 62.3% with a positive predictive value (PPV) of 72.9% and its specificity was 97.6% with a negative predictive value (NPV) of 96.2%. The area under the receiver operating curve (ROC) at 0.780 corresponded to a fair accuracy result (Table 3). The individual case definition categories showed lower sensitivities than their combined effect. Presence of two major risk factors was more sensitive at 33.3% sensitivity, followed by the category with one major and two minor factors and the least was the category with three minor factors. The areas under the ROC for the individual case categories were all lower than 0.7 showing their reduced accuracy.

 

Observed Gaps in the antenatal syphilis screening guidelines

 

ANC guidelines require first visit syphilis screening for pregnant women, however biomedical tests availability is a challenge due to; under-supply ( in all eight facilities), irregular submission of commodity consumption data for quantification of supply (two facilities) and inadequate resource allocation at district level to supplement supply; Although syndromic STI management is practiced in STI clinics, this is not practiced in antenatal clinic and antenatal attendees are left untreated in absence of biomedical tests; All ANC frontline staff were oriented on simple technique of using RST however they were not fully aware of revision in syphilis guidelines promoting their use.

 

 

This study confirmed that most of the pre-selected socio-demographic, behavioural and medical risk factors were significantly associated with maternal syphilis infection even in the study population validating their inclusion in the clinical assessment protocol. Respondents with syphilis were not only more likely to have co-morbid conditions like HIV and genital ulcer disease but also reported a history of multiple sexual partners, previous abortion, previous still birth delivery, previous syphilis infection and having a sexual partner with a sexually transmitted infection. The overall sensitivity of the protocol compared well to off-site field validation tests for point of care (POC) treponemal tests conducted in a syphilis clinic in Manaus, Brazil. In this study, off-site POC tests were reported to have sensitivities in the range of 45.8 to 66.7% [27]. The sensitivity of this study's assessment criteria however, with its inherent limitations was inferior to the widely recommended on-site rapid syphilis screening tests such as treponemal-based immuno-chromatographic strips (ICS). The latter have been reported to have field sensitivities ranging from 85 to 95% [28,29]. The possible sources of bias included information bias as it is unclear whether there were any risk factors relevant to the local study population omitted from the assessment criteria. Further, evaluation of clinical disease was essentially based on self-reported exposure to risk factors. It is likely therefore that the results may be affected by the participants' ability and willingness to recall and disclose exposure to certain risk behaviours. Generally, the clinical protocol performed reasonably well in predicting maternal syphilis infections even though the proportion of clinically presumed infections (8%) was lower than the actual sero-positive cases (9.3%). We considered the possibility that the observed protocol's performance might be due to sample-size related over-estimation of sero-prevalence which differed considerably from routine data in the study area. The 2007 Zambia Demographic Health Survey (ZDHS) estimated 4% prevalence levels among women of reproductive age group [30] and an observational study by Makasa et al showed declining national syphilis prevalence trends in concurrence to declines in HIV prevalence in Zambia [31].

However, this study's observed maternal syphilis sero-prevalence compares well to estimates from other studies in the country which generated information from antenatal clinics. In 2014 a study to evaluate rapid Dual HIV and syphilis tests, at the University Teaching Hospital in Lusaka, showed syphilis prevalence of 9.5% among women attending antenatal clinic [32]. This was similar to what was observed by Makasa et al, when they found high sero-prevalence of 10.8% in rural sites of Western province using antenatal sentinel surveillance data [31]. It is unclear although reasonable to assume that the disease-prediction performance of the clinical protocol would be affected by prevalence level of the disease. Therefore, additional and broader studies need to be performed to evaluate variability of the criteria's results at different syphilis point prevalence levels. The difficulty in predicting syphilis infections clinically with symptoms or risk factors is one of the main reasons WHO guided national policies recommend antenatal testing for all pregnant women using biomedical tests. However, despite having a policy in place, there are still challenges in ensuring that all ANC attendees access syphilis screening services. Some of these challenges arise from weaknesses in health systems such as; ineffective laboratory commodity supply and reporting systems, partial roll-out of the more cost-effective and easier to use RST and inadequate dissemination of revised screening guidelines. The overriding challenge affecting antenatal syphilis screening is the limitations in resources allocation to ensure availability of biomedical tests. The current level of commodity supply of biomedical syphilis test in the study area does not reflect political will to adhere to recommended policy of screening "all women attending antenatal clinic". Maintaining political will during implementation in Zambia is still a challenge despite recommendations which were accepted by the Ministry of Health to introduce point-of-care RST tests in national syphilis control guidelines [3]. This political will may diminish further as data show declining trends of syphilis prevalence [31]. There seems to be limited available alternatives to this problem. Some researchers have therefore recommended development of a dual test that would incorporate the much more politically acceptable HIV antenatal test [29] or epidemiological treatment for all pregnant women. This study's proposed clinical assessment protocol may be useful in identifying high risk infections for treatment. It could therefore be used for selective syphilis screening in declining disease-prevalence situations to limit costs of biomedical testing. It could also be used as an alternative screening tool in situations when biomedical tests are unavailable and in this situation it carries the advantage over epidemiological treatment in that sexual partners of the identified cases could also access treatment.

 

 

This study was able to illustrate that a clinical assessment protocol that is based on known socio-demographic, behavioral and medical risk factors of maternal syphilis can be used to identify women at high risk of infection. Despite its diagnostic limitations the protocol offers an alternative screening method, lacking in the national syphilis control guidelines, that could be used by frontline care providers. Even though Biomedical syphilis tests remain the most cost-effective means of identifying antenatal syphilis infections, there are some challenges related to health delivery systems in Zambia that have affected regular commodity availability. However, the proposed clinical protocol could offer an acceptable means of either identifying some cases in absence of biomedical tests or prioritizing those to be screened especially in resource limited settings.

 

 

The authors declare no competing interest.

 

 

JS developed the research proposal, conducting field work, initiating and writing the study report and manuscript. NC contributed in data analysis and manuscript writing. SN provided overall supervision for the research work, data analysis and write up this work.

 

 

Deepest gratitude is extended to the study participants, staff at the District Community Medical Offices and study facilities of Kaoma, Luampa and Nkeyema for their time, support and hard work during data collection. Special thanks goes to the Ministry of Community Development for the study's logistical support. The guidance provided by Professor Charles Michelo, departmental head, Public Health, School of Medicine at the University of Zambia was invaluable during the research process.

 

 

Table 1: univariate analysis: association between risk factors and maternal syphilis

Table 2: multivariate analysis: association of risk factors with maternal syphilis

Table 3: sensitivity, specificity and predictive value of proposed risk assessment criteria

 

 

1. World Health Organization. Investment case for eliminating mother-to-child transmission of syphilis. Promoting better maternal and child health and stronger health systems. Geneva, Switzerland. 2012. Google Scholar


2. Gomez GB, Kamb, ML, Newman JM, Broutet N, Hawkes S. Untreated maternal syphilis and adverse outcomes of pregnancy: a systematic review and meta-analysis. Bull World Health Organ. 2013 Mar 1; 91(3): 217-26. PubMed | Google Scholar


3. Ansbro EM, Gill MM, Reynold J, Shelley KD, Strasser S, Sripipatana T, Ncube AT, Mumba G, Peeling RW, Mabey D. Introduction of syphilis point-of-care tests, from pilot to National programme implementation in Zambia; a qualitative study of health care workers' perceptions on testing, training and quality assurance. PLoS One open access. June, 2015; 10(6): 18 pages. PubMed | Google Scholar


4. Maggwa BN, Askew I, Mugwe E, Hagembe B, Homan R. A case study of Nairobi City Council's Decentralized Syphilis Screening Programme in Antenatal Clinics. Population Council, Kenya, Nairobi, Kenya. 2001; p32. Google Scholar


5. World Health Organization. Prevention of mother-to-child transmission, Department of making pregnancy safer, World Health Organization. Geneva, Switzerland. 2007. Google Scholar


6. Zambian Ministry of Health. National programme for prevention of sexually transmitted infections: guidelines for use of rapid syphilis tests in Zambia. Lusaka, Zambia. 2011. Google Scholar


7. Rydzak CE and Goldie SJ. Cost-effectiveness of rapid point-of-care prenatal syphilis screening in sub-Saharan Africa. Sex Transm Dis. 2008 Sep; 35(9): 775-84. PubMed | Google Scholar


8. World Health Organization. Prevalence and incidence of selected sexually transmitted infections: Chlamydia, Neisseria gonorrhoeae, syphilis and Trichomonas vaginalis. Geneva, Switzerland. 2011. Google Scholar


9. Ministry of Health. National Reproductive Health Policy. Lusaka, Zambia. 2008. Google Scholar


10. Kaoma District Medical Office. Annual Health Management Information System Progress Reports. Ministry of Health. Kaoma, Zambia. 2012-2010. Google Scholar


11. Schackman BR, Neukermans CP, Fontain SNN, Nolte C, Joseph P, Pape J, Fitzgerald D. Cost-effectiveness of rapid syphilis screening in prenatal HIV testing programmes in Haiti. PLoS Med. 2007 May; 4(5): e183. PubMed | Google Scholar


12. Gloyd S and Mercer MA. Antenatal syphilis in sub-Saharan Africa: missed opportunities for mortality reduction. Health Policy Plan. 2001 Mar; 16(1): 29-34. PubMed | Google Scholar


13. Zambian Ministry of Health. Pregnancy, childbirth, postpartum and new-born care guideline: Agenda for essential practice in Zambia. Ministry of Health. Lusaka, Zambia. 2008. Google Scholar


14. Hawkes S, Miller S, Reichenbach L, Nayyar A, Buse K. Antenatal syphilis control: People, programmes, policies and politics. Bull World Health Organ. 2004 Jun; 82(6): 417-23. PubMed | Google Scholar


15. Schmid G. Economic and programmatic aspects of congenital syphilis prevention. Bulletin of the World Health Organization. June, 2004; 82(6): 417-423. PubMed | Google Scholar


16. Kebede E and Chamiso B. Prevalence of syphilis in pregnancy in Addis Ababa. East Afr Med J. April 2000; 77(4): 212-216. PubMed | Google Scholar


17. Walker DG and Walker GJA. Prevention of syphilis: Time for action. Bull World Health Organ. June 2004; 82(6): 401-402. PubMed | Google Scholar


18. Terris-Prestholt F, Vickerman P, Torres-Rueda S, Santesso N, Sweeney S, Malima P, Shelley KD, Gracia PJ, Bronzan R, Gill MM, Broutet N, Wi T, Watts C, Peeling RW, Newman L. The cost-effectiveness of 10 antenatal syphilis screening and treatment approaches in Peru, Tanzania and Zambia. Int J Gynaecol Obstet. 2015 Jun;130 Suppl 1:S73-80. PubMed | Google Scholar


19. Zhou H, Chen X-C, Hong F-C, Pan P, Yang F, Cai Y-M, Yin Y-P, Peeling RW, Mabey D. Risk factors for syphilis infection among pregnant women: results of a case-control study in Schenzhen, China. Sex Transm Infect. 2007 Oct; 83(6): 476-80. PubMed | Google Scholar


20. Miranda AE, Figueiredo NC, Pinto VM, Page K, Talhari S. Risk factors for syphilis in young women attending a family health programme in Vitoria, Brazil. Anais Brasileiros Dermatol. 2012; 87(1): 76-83. PubMed | Google Scholar


21. Todd J, Munguti K, Grosskurth H, Mngara J, Changalucha J, Mayaud P, Mosha F, Gayyole A, Mabey D, Hayes R. Risk factors for active syphilis and TPHA sero-conversion in a rural African population. Sex Transm Infect. 2001 Feb; 77(1): 37-45. PubMed | Google Scholar


22. Parker LA, Deschuter EJ, Bornay-Llinares FJ, Hernandez-Aguado I, Silva G, Piragine RC, Lumbreras B. Clinical and socioeconomic determinants of congenital syphilis in Posadas, Argentina. International Journal Infect Dis. April, 2012; 16(4): 256-261. PubMed | Google Scholar


23. Shah SA, Usman G, Ghazi A, Kristensen S, Sathiakumar N, Memon MA, Rubina J, Vermund SH. Prevalence of syphilis among antenatal clinic attendees in Karachi: Imperative to begin universal screening in Pakistan. J Pak Med Assoc. 2011 Oct; 61(10): 993-7. PubMed | Google Scholar


24. Uneke CJ, Ogbu O, Alo M, Ariom T. Syphilis serology in HIV-positive and HIV-negative Nigerians: the public health significance. Online J Health Allied Sci. 2006; 5: 1-8. PubMed | Google Scholar


25. Urassa WK, Kapiya SH, Msamanga GI, Antelman G, Coley J, Fawzi WW. Risk factors for syphilis among HIV-1 infected pregnant women in Da-er-salaam, Tanzania. African Journal Reprod Health. Dec, 2001; 5(3): 54-62. PubMed | Google Scholar


26. Nelson HD, Glass N, Huffman L, Villemyer K, Hamilton A, Frame P, Berg AO. Screening for syphilis; brief update. US Preventive Services Task Forces, Jul. 2004. Google Scholar


27. Benzakin AS, Sabido M, Galban E, Pedroza V, Arau jo AJG, Peeling RW, Mabey D. Field performance of a rapid point-of-care diagnostic test for antenatal syphilis screening in Amazon region, Brazil. Int J STD AIDS. 2011 Jan; 22(1): 15-8. PubMed | Google Scholar


28. Bonawitz RE, Duncan J, Hammond E, Nambule J, Sambambi K, Musonda V, Calise A, Knapp A, Mwale J, McCauley J, Thea D, Herlihy JM. Assessment of the impact of rapid syphilis tests on syphilis and treatment of pregnant women in Zambia. Int J Gynaecol Obstet. 2015 Jun; 130 Suppl 1: S58-62. PubMed | Google Scholar


29. Terris-Prestholt F, Watson-Jones D, Mugeye K, Kumaranayake L, Ndeki L, Weiss H, Changalucha J, Todd J, Lisekie F, Gumodoka B, Mabey D, Hayes R. Is antenatal syphilis screening still cost-effective in sub-Saharan Africa. Sex Transm Infect. 2003 Oct; 79(5): 375-81. PubMed | Google Scholar


30. Central Statistical Office, Ministry of Health, Tropical Diseases Research Centre, and University of Zambia and Macro International Inc. Zambia Demographic and Health Survey 2007. Central Statistical Office and Macro International. Claverton, Maryland. 2009. Google Scholar


31. Makasa M, Fylkesnes K, Michelo C, Kayeyi N, Chirwa B, Sandoy I. Declining syphilis trends in concurrence with HIV declines among pregnant women in Zambia: observations over 14 years of national surveillance. Sex Transm Dis. 2012 Mar; 39(3): 173-81. PubMed | Google Scholar


32. Yassa P, Hira R, Sibinda C, Katenga PT, Kim R, Seeda S, Mulenga V, Mukalay A, Glynis D, Chisela S, Tamba T, Tembo G. Evaluation of the Rapid dual HIV and Syphilis tests in women attending antenatal clinic at the University Teaching Hospital, Lusaka, Zambia. EMED Publications-International Infect Diseases. 2015; 1: 13. PubMed | Google Scholar