Share this article on:

Symptomatic HIV-Positive Persons in Rural Mozambique Who First Consult a Traditional Healer Have Delays in HIV Testing: A Cross-Sectional Study

Audet, Carolyn M. PhD*,†; Blevins, Meridith MSc; Rosenberg, Caitlin MA§; Farnsworth, Sarah BA§; Salato, José BA; Fernandez, Jorge BA; Vermund, Sten H. MD*,¶,‖

JAIDS Journal of Acquired Immune Deficiency Syndromes: August 1st, 2014 - Volume 66 - Issue 4 - p e80–e86
doi: 10.1097/QAI.0000000000000194
Implementation and Operational Research: Epidemiology and Prevention
IOR articles

Objective: Delays in HIV diagnosis and initiation of antiretroviral therapy are common even among symptomatic individuals in Africa. We hypothesized that antiretroviral therapy delays might be more common if traditional healers (THs) were the first practitioners consulted.

Design: Cross-sectional study.

Methods: We interviewed 530 newly diagnosed HIV-infected adults (≥18 years of age) who were clinically symptomatic at the time of HIV testing in 2 rural districts in Zambézia Province, Mozambique. We ascertained their previous health care seeking behavior, duration of their symptoms, CD4+ cell counts at the time of entry into care, and treatment provided by TH(s).

Results: Of 517 patients (97.5%) with complete histories, 62% sought care from a healer before presenting to the local health facility. The median time to first health facility visit from first relevant symptom was 2 months [interquartile range (IQR): 1–4.5] for persons who had not visited a healer, 3 months (IQR: 2–6) for persons visiting 1 healer, and 9 months (IQR: 5–12) for persons visiting >1 healer (P < 0.001). Healers diagnosed 56% of patients with a social or ancestral curse and treated 66% with subcutaneous herbal remedies. A nonsignificant trend toward lower CD4+ cells for persons who had seen multiple healers was noted.

Conclusions: Seeking initial care from healers was associated with delays in HIV testing among symptomatic HIV-seropositive persons. We had no CD4 evidence that sicker patients bypass THs, a potential inferential bias. Engaging THs in a therapeutic alliance may facilitate the earlier diagnosis of HIV/AIDS.

*Vanderbilt Institute for Global Health, Vanderbilt University School of Medicine, Nashville, TN;

Departments of Preventive Medicine;

Biostatistics, Vanderbilt University School of Medicine, Nashville, TN;

§Peace Corps, Mozambique, Quelimane and Maputo, Mozambique;

Friends in Global Health, Quelimane and Maputo, Mozambique; and

Department of Pediatrics, Vanderbilt University School of Medicine, Nashville, TN.

Correspondence to: Carolyn M. Audet, PhD, Department of Preventive Medicine, Vanderbilt University School of Medicine, Nashville, TN 37203 (e-mail:

Supported by the Centers for Disease Control and Prevention for data collection. C.M.A. is supported in part by Clinician and Translational Science Award/Vanderbilt Clinical & Translational Research Scholars Grant (KL2TR000446).

The Centers for Disease Control and Prevention reviewed the protocol and provided suggestions on sample size. The Clinician and Translational Science Award/Vanderbilt Clinical & Translational Research Scholars had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. C.M.A. had full access to all the data in the study and had final responsibility for the decision to submit for publication.

The remaining authors have no conflicts of interest to disclose.

Received January 15, 2014

Accepted April 02, 2014

Back to Top | Article Outline


People in sub-Saharan Africa often visit traditional healers (THs) before they seek care at an allopathic health facility.1–4 Unlike allopathic physicians, traditional and religious healers are thought by their clients to be able to diagnose ailments resulting from spiritual sources, such as spirits, sorcery, and social transgressions. There are different types of healers: healers can specialize in different types of treatment, including herbal-, spiritual-, and/or religious-based treatments.1,5–7 With the emergence of the HIV epidemic in Mozambique, the Ministry of Health [Ministério da Saúde de Moçambique (MISAU)] began encouraging patients to first seek care at a hospital; however, many Mozambicans continue to seek medical attention from healers before attending clinical services.4,8 In part, this is a consequence of cultural familiarity, availability, and inadequate allopathic health care.9 In Zambézia Province, there are an estimated 150 doctors and 850 nurses for >4 million persons (Fig. 1), a number dwarfed by the estimated 25,500 healers in the province (Ministry of Health, unpublished data, 2013). Although health care system limitations affect patient choices, many prefer visiting a healer regardless of the availability of clinical services.10



An estimated 12.6% of Zambézians aged 15–49 years were infected with HIV as of 2009.11 HIV care and services are located in district hospitals and a few other large clinical sites while rural areas provide counseling and testing services but refer patients to larger sites for care and treatment.12 Lower cadre health care providers, including medical technicians and nurses, provide nearly all of HIV care in the region.8

For a person living with HIV, the long-term effects of delayed, interrupted, or abandoned treatment are potentially severe, both because of higher rates of mortality and because of the risk of transmission to sexual partners due to high viral load.13,14 Persons with a CD4+ count of <200 cells per microliter are at far higher risk for opportunistic infections and malignancies; patients in programs that our group helps manage for MISAU had an initial median CD4 count of only 165 cells per microliter [interquartile range (IQR): 89–273] through the beginning of 2009. Earlier treatment (CD4 ≤ 550) has been linked to better health outcomes in the HIV Prevention Trials Network 052 trial; and the World Health Organization (WHO) now recommends antiretroviral therapy (ART) for persons with CD4 ≤500 for developing countries.15,16 Within ART-based HIV programs, healers have been documented as hindering prompt and effective provision of HIV-related health services (including testing),3,17,18 although this observation is not universal.19 Hence, earlier testing and successful linkage to ART-based care are vital to reducing morbidity and mortality in this rural region.

Few studies have attempted to quantify the time delay to testing associated with symptomatic HIV-positive individuals of unknown status who use THs before seeking allopathic services and receiving HIV testing.17,19 In the present study, we sought to: (1) estimate the length of time that elapses between onset of illness/symptoms in an undiagnosed HIV+ person to HIV testing among patients seeking clinical care at participating health facilities; (2) determine the impact of choosing a healer for initial care on delay to testing on the patients' CD+ cell count at the time of testing; and (3) assess what diagnoses healers are providing for their patients and whether they referred them for further testing at the health facility.

Back to Top | Article Outline


We conducted a cross-sectional study of newly diagnosed (past 3 months) HIV-positive adults (≥18 years) who were experiencing symptoms that caused them to seek allopathic health care services at the time of HIV testing. We interviewed 530 patients in 2 rural districts in southern Zambézia Province, Mozambique, between December 2011 and August 2012. Patients were eligible to participate if they: (1) received an HIV test at one of the eligible clinical sites; (2) had tested HIV positive in the past 3 months; (3) were 18 years of age or older; (4) were symptomatic before their HIV diagnosis; and (5) came to the hospital because of a physical complaint that led to referral for an HIV test. Of the 588 patients who were referred by nurses as being study eligible, 530 (90%) completed the interview. Interviews were conducted by study personnel who were not associated with the hospital staff. The interviews took 35–45 minutes and included questions about patient demographics, a retrospective structured interview to collect illnesses/conditions experienced continuously before HIV testing (all illnesses/conditions were recorded regardless of their connection with the eventual diagnosis of HIV and are hereinafter referred to as symptoms), length of time they had experienced this illness (we asked: “How long were you experiencing these symptoms before receiving an HIV test?”), and health-seeking behavior related to these symptoms. The patients' health identification number was also recorded to link with medical data, including CD4+ cell counts at first clinical visit and WHO status at treatment initiation. Patient's CD4 counts were collected from medical charts through January 2013. All patients should receive a CD4+ cell count and WHO status at treatment initiation, but reagent stock outs and clinician oversights result in only a minority of patients obtaining their CD4+ cell counts. Visits to THs were considered to have occurred for the purpose of the study at any time after initial symptoms began, and we also asked about the number of healers visited in that duration.

Back to Top | Article Outline

Ethical Approvals

This study was approved by the Vanderbilt University Institutional Review Board, the Comité Nacional de Bioética a Saúde in Mozambique, and the London School of Hygiene and Tropical Medicine.

Back to Top | Article Outline


Inhassunge and Namacurra districts are rural regions of Zambézia Province with estimated populations of 110,000 and 240,000, respectively. Care and treatment are accessible at 3 centers in Inhassunge and 7 centers in Namacurra (2 by mobile clinics). Educational attainment in the region is typically less than primary school and few people have any monthly income.4,10,20 Median CD4+ cell counts at treatment initiation in our President's Emergency Plan for AIDS Relief–supported program in Mozambique have risen by <10 cells per microliter over the past 4 years, despite substantial expansion of HIV testing.21

Back to Top | Article Outline

Study Size

We used data from a previously conducted cross-sectional study to estimate healer use in Zambézia4 (assuming 30% of patients had visited a TH for their symptoms) and data from a study in West Africa to estimate variability (SD = 80 days) in delays to HIV testing among symptomatic HIV-positive people17 to inform sample size requirements. We aimed to enroll 410 HIV-positive individuals, assuming a mean difference in delay time of 28 days, to be able to reject the null hypothesis that the mean delay time of healer users and nonusers are equal with probability (power) 0.9 and type I error of 0.05; the actual enrollment was 530 (ie, 129% of the target sample size).

Back to Top | Article Outline

Statistical Analysis

Summary statistics, including frequency/percentage (for categorical data) and median/IQR (for continuous data), are provided for sociodemographic data, survey responses, and clinical outcomes, stratified by use of traditional medicine. Patients who saw a TH any time after initial symptoms began were asked whether they saw more than 1 healer, and these data served as 2 primary outcomes: (1) visitation of a TH and (2) number of healers consulted. Groups were compared using χ2 tests for categorical data and rank sum tests for continuous data. Scatterplots for number of healers vs. delay time or CD4 count were fit with Loess curves, and Spearman correlation coefficients were estimated.22 Time from symptom to testing was modeled using multivariable linear regression with log-transformed delay time to testing as the outcome, including district, age, sex, education, travel time to clinic, transportation to clinic, consultation with religious leader, and care from healer as covariates. Multiple imputation was used to account for missing values of covariates and to prevent case-wise deletion of missing data; 478 (92.4%) patients had complete data for all covariates. To relax linearity assumptions, we modeled age using a restricted cubic spline with 3 knots.23 All hypothesis testing was 2 sided with a level of significance set at 0.05. We used R-software 2.15.1 ( for all data analyses. Analysis scripts are available at

Back to Top | Article Outline


Of the 530 participants who completed the study, 517 (97.5%) provided responses to the question “How long were you experiencing these symptoms before receiving an HIV test?” and were therefore included in the analysis (Table 1). Participants were 55% female, with a median age of 29 years (IQR: 23–37), median education of 4 years (IQR: 0–6), and living a median of 60 minutes (IQR: 30–90) from the hospital where they were being treated. Only 19% of the participants spoke Portuguese in their homes (Echuabo is the local language of preference). Self-identified religious identity was 54% Catholic, 20% Muslim, 9% Seventh Day Adventist, 7% Baptist or Assembly of God, 5% other Christian, and 5% none.



A majority of participants (n = 321; 62%) first sought care from a healer before seeking care from the health facility. Compared with persons who did not first visit a healer, participants who first visited a healer for their symptoms were more likely to be female, had a median 1 year less of formal education (3 vs. 4 years), were more likely to have to use a car/motorcycle to get to the clinic (vs. walking or bicycle), and were more likely to have also sought care from a religious leader. Median time to clinical visit from first symptom(s) was 2 months (IQR: 1–4.5) for those who did not visit a healer, 3 months (IQR: 2–6) for those who saw a single healer, and 9 months (IQR: 5–12) among those who saw 2 or more healers (P < 0.001). A scatterplot with Loess curve of delay time by number of healers visited shows a trend of increasing delay times among those who consulted with increasing numbers of healers (Fig. 2A) (Spearman ρ = 0.54; P < 0.001). A scatterplot with Loess curve of CD4 cell counts shows an insignificant decrease in CD4 by number of healers (Fig. 2B) (Spearman ρ = −0.05; P = 0.49).



We conducted a linear regression model to determine factors associated with delayed time to clinical visit (Table 2). Covariates identified a priori included sex, age, education, travel time to clinic, transportation, number of consultations with a healer before the clinic (0 to >4), consultation with a religious leader, and community (Inhassunge or Namacurra). Female patients had an average 30% longer delay to testing than males (P = 0.02). A patient aged 20 years had 30% shorter delay to testing compared with a patient aged 30 years (P = 0.02). We failed to detect independent associations between education, travel time, consultation with religious leader, and transportation with delay to testing. Patients in Namacurra had 2.0 times higher average delayed times to clinical visit than those at Inhassunge (P < 0.001).



Previous consultation with a healer was associated with delayed time of diagnosis (P < 0.001). Compared with someone who did not consult any healer, participants who visited a single healer had a 1.5 times longer delay, those who consulted with 3 healers had a 2.1 times longer delay, and those who consulted with 5 or more healers had a 5.7 times longer delay to testing. When running the same multivariable model dichotomizing any use of healers as the main exposure of interest, people who consulted a healer had a 2.4 times longer delay than those who did not {effect estimate [95% confidence interval (CI)]: 2.42 (1.91 to 3.05); data not shown}. From the model, we estimated mean delay time by healer use for our average participant: a woman, age 29 years, with 4 years of formal education, who walks to the clinic in 1 hour, lives in Namacurra, and does not consult with a religious leader on health concerns; that patient had a mean delay of 99 days (95% CI: 74 to 132 days) from first symptom to testing if she did not visit a healer (Table 3). Her delay time increased as she consulted with greater numbers of healers, such that this patient was delayed an average of 1.5 years (95% CI: 415 to 767 days) from first symptom to testing if she sought care from 5 or more healers (Table 3).



Patient symptoms were documented at the time of enrollment (data not shown). Pulmonary symptoms (coughing, trouble breathing), gastrointestinal issues (diarrhea), fever, musculoskeletal problems (body aches), and neurologic/psychological complaints (dizziness, headaches) were the most common symptoms reported. People who first visited a healer were more likely to report having pulmonary, musculoskeletal, systemic problems (including skin rashes), and genitourinary system/pregnancy-related symptoms than those who first sought care from a clinical site. Specifically, patients with cough (P = 0.005), body aches (P = 0.004), diarrhea (P = 0.003), muscle weakness (P = 0.04), and abdominal pains (P = 0.006) were more likely to first seek care from a TH. Self-reported delay time was longer among people who first saw a healer who experienced abdominal pain (157 vs. 269 days; P < 0.001), lack of appetite (122 vs. 240 days, P = 0.01), and those who experienced vaginal discharge (224 vs. 426 days, P < 0.02).

Of patients seen by just 1 healer, 159 (55%) of 323 patients were administered herbs, “injected” subcutaneously using razor blades, and rubbing the herbs into the cuts. Of 188 persons seen by ≥2 healers, 66% reported having had the subcutaneous herbal inoculation. Although 12 training sessions had been conducted with THs in these 2 communities by MISAU and partners to help healers identify the symptoms of HIV, only 30 (9%) of our participants reported being referred to the health facility by a healer. Three diagnoses were most commonly reported to have been made by healers: being cursed (182, 56%), of biological origin (49, 15%), and angering spirits (31, 10%). Of those 30 patients who reported healer referral to the health facilities for treatment, 90% reported this as their primary reason for their presentation at the health facility.

Back to Top | Article Outline


We found significant differences in delayed time to testing among those who did and did not visit a healer before visiting a clinic for their symptoms. If we can effectively partner with healers to enable early referral in a given community, we may be able to facilitate earlier HIV testing, enabling needed earlier therapy.24 Patient referral from a TH to clinical site is fairly straightforward: healers provided patients with a referral form indicating their symptoms; patients bring this form to the clinical site where they present it to the intake nurse and wait in line as would any other patient. Our previous studies revealed that patients who admitted to visiting a healer were often criticized by clinicians in Mozambique.25 New training and clinical practices are aiming to change this dynamic: to encourage patients to accept healer referral, the consultation fee (1 MT or US $0.03) has recently been waived for these patients and in some sites the patient is allowed to “jump the queue” and see the clinicians faster than patients who were not referred by a healer.

We saw a greater proportion of women seeking care from a healer. We can only hypothesize the reasons for this difference: (1) women have lower socioeconomic status and are required to complete a disproportionate amount of farming labor and housework in these communities which may limit their ability to travel to clinical sites for treatment4,20 and (2) women may prefer to seek care from a TH for linguistic reasons: fewer women than men in these communities speak Portuguese, the language of preference at health facilities.20,25

The use of healers as primary care providers in rural Mozambique suggests the importance of engaging healers in HIV testing strategies.10 However, the paltry 9% of patients referred for clinical treatment, despite prior training of 200 healers in these 2 districts, is disappointing. The educational training that we have already done has not been effective to motivate referral of ill patients to health facilities for HIV testing. Recent examples of successful engagement of healers and allopathic providers may provide encouragement for moving forward with the goal of linking the 2 systems further.26–31 Healers need to have a clearly defined role in patient care that extends beyond referral, such as adherence counseling, and ideally would be compensated for worthwhile services, whether financially or otherwise. Once referred to the hospital, patients encounter an additional challenge: a 2012 study in Zambézia found that even when healers referred patients to the health facility, the proportion of patients tested for HIV at first visit was only 3%.32 Further investigation is needed to determine whether this suggests a lack of respect from allopathic providers for the opinions of healers or reflects a general failure to test symptomatic patients. Following an HIV-positive diagnosis, the course of traditional treatments provided is not well understood.33 Future research in this area is required to ensure that healer treatments do not encourage the co-administration of potentially hepatotoxic and/or nephrotoxic agents or result in additional possible deleterious drug–herb interactions.34–38

Treatment strategies used by healers to treat patient symptoms put themselves and their patients at risk of blood-borne pathogens.10 Healers who practice cutting and herbal rubbing on their patients increase direct exposure to infected blood by reusing razors and other implements such as sticks for rubbing the herbs into the skin. A paucity of rubber gloves and nonreusable sterile razors suggest that the risk of interpatient and/or healer HIV/hepatitis transmission may be substantial.39

Study strengths include the large sample size and the high rate of participation, suggesting potential generalizability to this rural population. The collection of clinical data to compare with self-reported time delays allowed us to validate recall, but we were limited by the availability of CD4+ cell counts in only a subset of subjects. This percentage of missing CD4+ cell data (39%, Table 1) is the same as found among the HIV patient population in these rural Mozambican clinics (40%, data not shown). Although we have no evidence of nonrandomness of ascertainment, we cannot be sure that the CD4 counts are representative of the full population, a limitation in our study. Because CD4 counts were lower in persons who had seen more than 1 TH, we do not believe that sicker persons were bypassing healers to go directly for care. As a retrospective cross-sectional study, patient recall of time since symptoms began can be incorrect. If there were systematic bias in the data by use of traditional medicine, our results will not accurately reflect true delays to clinical visits. We did not capture data on length of time patients had been experiencing symptoms before seeking care from a TH of why they chose to visit a healer vs. a clinician, but this is an important area for future research. As our study was limited only to people who have received an HIV-positive test result, we did not capture data about those who had HIV but refused to test or after testing did not present at the clinical site for test results. Persons who might have died of HIV before reaching the HIV clinics will not have presented to us for the healer assessment, an unseen cohort whose characteristics of advanced HIV disease and failure to present to the clinic might indicate our findings to be a minimum estimate of the true delays associated with healer visits.40 Also, as our data are only from 2 rural sites, the data may not be generalizable to urban areas of Mozambique or to other sub-Saharan African countries.

Our findings can alert clinical HIV service providers in Africa that initially seeking care from healers was frequent and was associated with delays in HIV testing, even among symptomatic HIV-positive patients. We cannot be sure that this is causal, but even the association is of interest, given the low CD4+ cell counts seen all over Africa in persons presenting for HIV care.41,42 With the majority of our patients first seeking care from a healer when sick, working with healers to facilitate a quick diagnosis and referral to the clinic when appropriate would seem to be a priority in HIV programs in Africa. The 9% proportion of healer referrals to the health facilities was discouraging given prior HIV training that had been provided to healers in these districts. As healers continue to neglect patient referral even after receipt of information about available services, education alone is likely not sufficient; a new model to incentivize referral behavior is needed. In addition, health care workers must be encouraged to welcome such referrals, not always a guaranteed reaction in the setting of hostility between healers and allopathic providers. Development and testing of a model that incentivizes healer referral of sick patients for HIV testing and care will be a high priority for new research, as will determining a productive long-term relationship for augmenting treatment adherence.

Back to Top | Article Outline


The authors would like to thank the participants who participated in the study, the Mozambique Ministry of Health, and the clinicians at the 2 study clinics.

Back to Top | Article Outline


1. Kale R. Traditional healers in South Africa: a parallel health care system. BMJ. 1995;310:1182–1185.
2. Stekelenburg J, Jager BE, Kolk PR, et al.. Health care seeking behaviour and utilisation of traditional healers in Kalabo, Zambia. Health Policy. 2005;71:67–81.
3. Banda Y, Chapman V, Goldenberg RL, et al.. Use of traditional medicine among pregnant women in Lusaka, Zambia. J Altern Complement Med. 2007;13:123–127.
4. Audet CM, Sidat M, Blevins M, et al.. Health seeking behavior in Zambezia Province, Mozambique. SAHARA J. 2012;9:41–46.
5. Green EC. Traditional healers and AIDS in Uganda. J Altern Complement Med. 2000;6:1–2.
6. Green EC, Zokwe B, Dupree JD. The experience of an AIDS prevention program focused on South African traditional healers. Soc Sci Med. 1995;40:503–515.
7. Pfeiffer J. Commodity fetichismo, the Holy Spirit, and the turn to Pentecostal and African Independent Churches in Central Mozambique. Cult Med Psychiatry. 2005;29:255–283.
8. Audet CM, Burlison J, Moon TD, et al.. Sociocultural and epidemiological aspects of HIV/AIDS in Mozambique. BMC Int Health Hum Rights. 2010;10:15.
9. Audet CM, Groh KE, Moon TD, et al.. Poor quality health services and lack of program support leads to low uptake of HIV testing in rural Mozambique. J Afr AIDS Res. 2012;4:327–335.
10. Audet CM, Blevins M, Moon TD, et al.. HIV/AIDS-related attitudes and practices among traditional healers in Zambezia Province, Mozambique. J Altern Complement Med. 2012;18:1–9.
11. INSIDA. National survey on prevalence, behavioral risks and information about HIV and AIDS (2009 INSIDA). 2009. Available at: Accessed August 14, 2013.
12. Moon T, Burlison J, Sidat M, et al.. Lessons learned while implementing an HIV/AIDS care and treatment program in rural Mozambique. Retrovirology (Auckl). 2010;3:1–14.
13. Chi BH, Stringer JS, Moodley D. Antiretroviral drug regimens to prevent mother-to-child transmission of HIV: a review of scientific, program, and policy advances for sub-Saharan Africa. Curr HIV/AIDS Rep. 2013;10:124–133.
14. Baggaley RF, White RG, Hollingsworth TD, et al.. Heterosexual HIV-1 infectiousness and antiretroviral use: systematic review of prospective studies of discordant couples. Epidemiology. 2013;24:110–121.
15. Cohen MS, Chen YQ, McCauley M, et al.. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med. 2011;365:493–505.
16. World Health Organization. WHO Consolidated Guidelines on the Use of ARV Drugs for Treating & Preventing HIV Infection. Geneva: WHO Press; 2013:273. Available at:
17. Okome-Nkoumou M, Okome-Miame F, Kendjo E, et al.. Delay between first HIV-related symptoms and diagnosis of HIV infection in patients attending the internal medicine department of the Fondation Jeanne Ebori (FJE), Libreville, Gabon. HIV Clin Trials. 2005;6:38–42.
18. Smyth A, Martin M, Cairns J. South Africa's health. Traditional healers may cause dangerous delays. BMJ. 1995;311:948.
19. Horwitz RH, Tsai AC, Maling S, et al.. No association found between traditional healer use and delayed antiretroviral initiation in rural Uganda. AIDS Behav. 2012.
20. Vergara A, Blevins M, Vaz LM, et al.. Baseline Survey Report: Phase I and II: Zambezia Wide. Nashville, TN: Vanderbilt Universit; 2011.
21. Moon TD, Vaz LM, Parrish DD, et al.. Predictors of adult mortality and loss-to-follow-up within two years of initiation of combination antiretroviral therapy in rural Mozambique. Poster presented at: XIX International AIDS Conference; July 22–27, 2012; Washington, DC.
22. Cleveland WS, Grosse E, Shyu WM. Local regression models. In: Chambers JM, Hastie TJ, eds. Statistical Models in S. Pacific Grove, CA: Wadsworth & Brooks/Cole; 1992.
23. Harrell FE. Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. New York, NY: Springer; 2001.
24. Severe P, Juste MA, Ambroise A, et al.. Early versus standard antiretroviral therapy for HIV-infected adults in Haiti. N Engl J Med. 2010;363:257–265.
25. Groh K, Audet CM, Baptista A, et al.. Barriers to antiretroviral therapy adherence in rural Mozambique. BMC Public Health. 2011;11:650.
26. Ssali A, Butler LM, Kabatesi D, et al.. Traditional healers for HIV/AIDS prevention and family planning, Kiboga District, Uganda: evaluation of a program to improve practices. AIDS Behav. 2005;9:485–493.
27. Green EC. The participation of Africa traditional healers in AIDS/STI prevention programmes. AIDS Link. 1995;36:14–15.
28. Makundi EA, Malebo HM, Mhame P, et al.. Role of traditional healers in the management of severe malaria among children below five years of age: the case of Kilosa and Handeni Districts, Tanzania. Malar J. 2006;5:58.
29. Mashamba T, Peltzer K, Maluleke TX, et al.. A controlled study of an HIV/AIDS/STI/TB intervention with faith healers in Vhembe District, South Africa. Afr J Tradit Complement Altern Med. 2011;8(5 suppl):83–89.
30. Mbeh GN, Edwards R, Ngufor G, et al.. Traditional healers and diabetes: results from a pilot project to train traditional healers to provide health education and appropriate health care practices for diabetes patients in Cameroon. Glob Health Promotion. 2010;17(2 suppl):17–26.
31. Peltzer K, Mngqundaniso N, Petros G. A controlled study of an HIV/AIDS/STI/TB intervention with traditional healers in KwaZulu-Natal, South Africa. AIDS Behav. 2006;10:683–690.
32. Audet CM, Salato J, Blevins M, et al.. Educational intervention increased referrals to allopathic care by traditional healers in three high HIV-prevalence rural districts in Mozambique. PLoS One. 2013;8:e70326.
33. Peltzer K, Friend-du Preez N, Ramlagan S, et al.. Traditional complementary and alternative medicine and antiretroviral treatment adherence among HIV patients in Kwazulu-Natal, South Africa. Afr J Tradit Complement Altern Med. 2010;7:125–137.
34. Mills E, Cooper C, Seely D, et al.. African herbal medicines in the treatment of HIV: Hypoxis and Sutherlandia. An overview of evidence and pharmacology. Nutr J. 2005;4:19.
35. Babb DA, Pemba L, Seatlanyane P, et al.. Use of traditional medicine by HIV-infected individuals in South Africa in the era of antiretroviral therapy. Psychol Health Med. 2007;12:314–320.
36. Langlois-Klassen D, Kipp W, Jhangri GS, et al.. Use of traditional herbal medicine by AIDS patients in Kabarole District, western Uganda. Am J Trop Med Hyg. 2007;77:757–763.
37. Langlois-Klassen D, Kipp W, Rubaale T. Who's talking? Communication between health providers and HIV-infected adults related to herbal medicine for AIDS treatment in western Uganda. Soc Sci Med. 2008;67:165–176.
38. Cordier W, Steenkamp V. Drug interactions in African herbal remedies. Drug Metabol Drug Interact. 2011;26:53–63.
39. Tjøtta E, Hungnes O, Grinde B. Survival of HIV-1 activity after disinfection, temperature and pH changes, or drying. J Med Virol. 1991;35:223–227.
40. Hoover DR, Muñoz A, Carey V, et al.. The unseen sample in cohort studies: estimation of its size and effect. Multicenter AIDS Cohort Study. Stat Med. 1991;10:1993–2003.
41. Gupta A, Nadkarni G, Yang WT, et al.. Early mortality in adults initiating antiretroviral therapy (ART) in low- and middle-income countries (LMIC): a systematic review and meta-analysis. PLoS One. 2011;6:e28691.
42. Moon TD, Burlison JR, Blevins M, et al.. Enrolment and programmatic trends and predictors of antiretroviral therapy initiation from President's Emergency Plan for AIDS Relief (PEPFAR)-supported public HIV care and treatment sites in rural Mozambique. Int J STD AIDS. 2011;22:621–627.

HIV/AIDS; Mozambique; traditional healers; testing delays; health-seeking behavior; sub-Saharan Africa

© 2014 by Lippincott Williams & Wilkins