Publication Date:
01 May 2014
Citation:
Gloyd SS, Robinson J, Dali SA, Granato SA, Bartlein R, Kouyate S, et al. (2014). PMTCT cascade analysis in Côte d’Ivoire: Results from a national representative sample. HIVCore Final Report: USAID.
Abstract
Without treatment, an HIV-positive pregnant woman has a 30 to 35 percent chance of transmitting the infection to her child. Early detection and appropriate treatment during pregnancy, childbirth, and breastfeeding can reduce those odds to less than 5 percent. Consequently, the World Health Organization (WHO) and countries around the world have instituted prevention of mother-to-child transmission of HIV (PMTCT) programs, consisting of the following “PMTCT cascade”:
- Testing for HIV at the first antenatal visit (ANC1).
- CD4 test of HIV-positive patients (optional, depending on PMTCT regimen).
- Antiretroviral (ARV) prophylaxis to mother throughout pregnancy and breastfeeding.
- ARV prophylaxis to child at delivery and throughout breastfeeding.
- HIV testing of child and immediate initiation of antiretroviral therapy (ART) for children found to be HIV infected.
In Côte d’Ivoire, as in much of Africa, PMTCT programs have performed less than optimally due to bottlenecks throughout the PMTCT cascade. These bottlenecks often cause delays in initiating care and contribute to loss to follow up (LTFU) in one or more stages of the PMTCT cascade. Understanding each step of the cascade, and the role that health system and community factors play in enabling or mitigating PMTCT bottlenecks at each step, can contribute to an increase in effective PMTCT outcomes.To help inform the 2012 rollout of the WHO “Option B” treatment regimen1 in Côte d’Ivoire, an assessment of a nationally representative sample of existing PMTCT programs in Côte d’Ivoire was planned and carried out by a consortium including Côte d’Ivoire’s Ministère de la Santé et de la Lutte contre le Sida (MSLS), the Université Allasane Ouattara, the University of Washington (UW), and Health Alliance International (HAI). The study was conducted under the U.S. Agency for International Development’s (USAID) HIVCore project led by the Population Council, with additional financial and technical support from the USAID office in Côte d’Ivoire. The objectives of the study were to identify bottlenecks and time delays in the PMTCT cascade, and suggest recommendations to improve current PMTCT services to optimize the impact of Option B in Côte d’Ivoire. It was hoped that findings would stimulate discussion of testable interventions to improve the PMTCT cascade as Option B is rolled out.
METHODOLOGY
Drawing from a base of 734 health facilities providing PMTCT services in Côte d’Ivoire, this study utilized primary and secondary data from a randomly selected nationwide sample of 30 sites that had served at least 10 HIV-positive patients in 2011. At each of the 30 sites, quantitative data were abstracted from available health facility registries for the period June 2011 to May 2012 and a sample of up to 20 patient charts of HIV-positive women (n=330), as well as from a sample of up to 20 mother-child vaccination booklets (carnets2) (n=590) per site. Qualitative data, including semi-structured interviews with key informants and direct observations, were recorded by seven two-person study teams over the course of two days at each facility. Data collection was carried out over a two-week period in March 2013.
KEY FINDINGS
Registry data at the 30 sites were frequently inconsistent. Availability of standard registries was inconsistent across the 30-site sample. On average, sites recorded and stored data related to PMTCT in 11 different registries, reports, or other unofficial data sources. Using all available data sources at the 30 study sites, 11 registry indicators were assessed. Of the 30 study sites, only five had complete registry data available for the study time period. One site had no data available for seven indicators and was excluded from registry-based analysis. Seven sites had incomplete and/or missing data on 6–10 indicators, and the remaining 17 sites had missing or incomplete data on 1–5 indicators. When on-site registry indicators were compared against the same indicators in the national database, inconsistencies were found in every indicator and at every site. On average, more than half of all indicators compared had a discrepancy of larger than 5 percent between on-site data and the same data in the national database.
Clinics demonstrated significant losses and delays throughout the PMTCT cascade.Pooling HIV-positive pregnant women as the denominator across the 29 sites, the following rates were observed for each step of the PMTCT cascade: delivery of ARV prophylaxis and/or combination ART to the mother–77 percent; delivery of ARV prophylaxis intended for the newborn–70 percent; administration of first infant HIV test–50 percent; mother’s CD4 count taken–43 percent; initiation of cART after CD4 testing–16 percent. Comparing sites located in Abidjan to those outside Abidjan, the only indicator that showed a notable difference was delivery of ARV prophylaxis intended for the newborn (77 percent vs. 60 percent respectively; two-tailed t-test assuming unequal variances gave p-value of .0308). Using the 2011 DHS, we estimated a 91 percent antenatal care (ANC) attendance rate among pregnant women in Côte d’Ivoire. We used registry and carnetdata to estimate that in all of Côte d’Ivoire: approximately 76 percent of HIVpositive pregnant women are tested for HIV in ANC, approximately 58 percent receive prophylaxis or cART during pregnancy, and only about 38 percent have their newborn tested for HIV within six weeks of delivery. Additionally, we estimated that only 32 percent of HIV-positive pregnant women in Côte d’Ivoire have a CD4 count taken to determine eligibility for cART. However, the caveat remains that the registry and carnet data were inconsistent as stated above.
Nearly one-third of sites did not create charts for HIV-positive patients. National standards in Côte d’Ivoire require opening a clinical chart for PMTCT immediately after a patient receives a positive HIV test. We found that nearly one third of the study sites did not follow these standards. These sites reported that typically patient charts are opened when CD4 counts or clinical staging indicated eligibility for cART. This skewed our chart sample toward patients on cART, as HIV-positive patients who had not yet started cART might not have a chart. Workforce patterns, not size and location of facility, had a significant association with PMTCT performance.Sites were assessed for overall PMTCT performance using a composite score based on monthly rates for child prophylaxis, cART initiation, and newborn HIV testing. The 10 sites with the lowest PMTCT scores were compared with the 10 sites with the highest PMTCT scores. On average (mean), high-performing PMTCT sites had more skilled personnel and personnel trained in PMTCT-related services per 1,000 ANC1 than did low–performing PMTCT sites, including doctors (mean among high performers (MH)=7/1,000; mean among low performers (ML)=2/1,000; p=.030), nurses (MH=7/1,000; ML=2/1,000; p=.043), and laboratory technicians (MH=2/1,000; ML=1/1,000; p=.021) as well as numbers of staff trained in PMTCT (MH=11/1,000; ML=4/1,000; p=.005), and staff engaged in ART follow-up (MH=8/1,000; ML=2/1,000; p=.009). Size and location of facility did not have a significant association with PMTCT performance score, while workforce patterns were significantly associated with PMTCT performance.Patterns of patient flow and health care environment in the study sites varied considerably.Results from interviews andobservations demonstrated that the patterns of patient flow as measured by the number and order of steps through PMTCT services in the study sites and the health care environment, varied considerably. Long wait times, insufficient infrastructure, lack of medical and office equipment and tools, inability to ensure confidentiality, and poor ventilation were common barriers expressed by key informants and observed by interview teams. Additionally, the key informants’ estimates of time between ANC1 and the return of CD4 results to the patient significantly differed between high-performing PMTCT sites and low-performing PMTCT sites (MH=5.1 days; ML=11.4 days; p=.001); however, reported time elapsed derived from patient charts revealed actual median delays of 19.5 and 31 days respectively. In most sites, health care worker dedication and engagement were seen as facilitators of patient retention.
CONCLUSIONS AND RECOMMENDATIONS
This nationally representative sample of antenatal clinics performing PMTCT in Côte d’Ivoire demonstrated widespread problems in infrastructure, patient flow, and quality of care. Complete and accurate record-keeping was the exception rather than the norm. Significant losses and delays were observed throughout the PMTCT cascade, including poor retention for prophylaxis (both mothers and newborns), PCR testing, and CD4 testing. The higher performing sites had significantly greater numbers of skilled personnel and PMTCT training of skilled health personnel than the lower performing sites.
The burden and confusing nature of so many registries as well as extremely high levels of data inconsistencies between on-site data and national data suggest that monitoring and evaluation should be greatly simplified. More focused attention to fewer indicators, fewer reports, and fewer registries would likely improve their usefulness in monitoring and evaluation. In addition, a system for cross-checking issues of on-site adherence to national protocols (both clinical and administrative), availability of necessary supplies, and effective communication (supervision and feedback) between district and facility levels should be established. The results of this assessment suggest that antenatal care clinics in Côte d’Ivoire may struggle with the national rollout of Option B, where adequate assessment, medication, and follow up of every HIV-positive pregnant woman will be necessary for effective results. A systematic focus on patient follow up with adequate charting and communication with patients is critical to success. A robust evaluation of the rollout is also recommended. We expect bottlenecks and challenges will likely occur, which will need to be identified and addressed to maximize the efficiency and impact of this critical intervention.