Screening tool for HIV in children not being used effectively by health care workers

Incorrect use in routine practice of a World Health Organization (WHO)/UNICEF HIV screening tool for children at primary health care clinics in Limpopo and KwaZulu Natal provinces, South Africa, leads to the failure of life-saving interventions, Christiane Horwood and colleagues reported in a study in the September 22 2009 edition of BMC Pediatrics. Forty per cent of trained health workers failed to identify HIV in any child and not one was able to classify every child correctly for HIV. In South Africa where HIV prevalence rates among pregnant women remain at 29%, the burden of paediatric HIV disease continues to grow. Limpopo – with a mostly rural population of 5.5 million – has high rates of poverty and poor access to basic services. While KwaZulu Natal has less poverty, half of the 10-million-strong population lives in rural areas and antenatal HIV prevalence in 2006 was close to 40%. Insufficient testing and follow-up of HIV-exposed children leads to high mortality rates with over half of untreated children dying within the first two years of life. Few children who need antiretroviral treatment receive it in spite of it being free. Improved follow-up of HIV exposed children, increased early identification of children with symptomatic HIV and improved access to ART for children are urgently needed. New guidelines from WHO recommend that, where virological testing is unavailable, children should be started on antiretroviral treatment based on clinical diagnosis alone followed by quick confirmation of HIV status. WHO and UNICEF developed the Integrated Management of Childhood Illnesses (IMCI) strategy to improve child survival in resource-poor settings. Focusing on the well-being of the whole child, the aim is to reduce death, illness, disability, and to promote improved growth and development among children under five years of age. South Africa adopted these guidelines as the standard of care for children at the primary level in 1997. Multi-country evaluations of the IMCI strategy indicate, when used correctly, improved health worker performance and quality of care as well as a reduction in under-five mortality and improved nutritional status. The guidelines have been adapted to incorporate a validated HIV component (including an algorithm) to identify and manage HIV-infected (at risk for early death) and exposed (symptomatic) children. The IMCI course includes comprehensive training on this component. For effective use of the algorithm, healthcare workers are expected to ask every mother bringing a sick child to a healthcare facility whether she has been tested for HIV, so that children may be classified as HIV-exposed, and all children should be assessed for clinical symptoms suggestive of HIV. The presence of three or more symptoms should trigger further investigation and the carer should be advised of the need for the child to be tested for HIV. In this first known evaluation of the IMCI/HIV guidelines, the study was designed to show how the guidelines are used by IMCI trained health workers, the validity of the HIV algorithm when used by expert IMCI practitioners in routine practice and the burden of HIV disease among under-fives attending primary health care facilities in Limpopo and KwaZulu Natal provinces. Between May 2006 and January 2007 seventy-seven randomly selected IMCI trained health workers were observed by IMCI experts in 74 primary health care facilities in Limpopo and KwaZulu Natal provinces. All sick children between the ages of two months and five years were eligible. Consultations with a total of 1357 sick children were observed. A different IMCI expert reassessed each child to confirm correct findings. Consent for HIV testing for all children who attended was requested from parents or legal guardians. Positive rapid tests were confirmed with HIV polymerase chain reaction (PCR) tests in children under 18 months of age. HIV-positive children had CD4 counts and HIV clinical staging done. Each health worker was observed for a mean of 2.2 days and 17.7 consultations. The average age of the observed children was 19.6 months, of whom 40.7% (552) were under one year of age. A third of all consultations were observed in Limpopo and the remaining two-thirds in KwaZulu Natal. Of the 1064 children with available HIV test results, 76 tested positive giving an HIV prevalence rate of 7.1% (CI: 5.7 to 8.9%) among children in primary healthcare clinics. Of these 76, one was on antiretroviral treatment. Following CD4 counts or, if unavailable, WHO clinical staging, ART was indicated for 84% (63 of 75) of the remaining children. When compared to the HIV test results, IMCI experts skilled at using the HIV algorithm correctly identified 90.8% (69 of 76) of HIV-infected children as either suspected symptomatic HIV or HIV-exposed and therefore in need of further investigation. This shows that when used correctly the HIV algorithm is an effective screening tool and can lead to improved access to life-saving treatment for HIV infected and exposed children. In comparison, over 40% of IMCI-trained health workers failed to identify HIV in any child because of poor or incomplete use of the HIV component. And nine did not classify the disease stage of any child with HIV correctly. Even when health workers classified children with suspected symptomatic HIV, the need for testing, cotrimoxazole prophylaxis and feeding advice was only communicated to 64%, 31% and 43% of carers, respectively. The authors suggest several reasons for poor use of the algorithm:
Inadequate training
Lack of a clear understanding that the HIV algorithm is a screening tool and not a diagnostic test. IMCI training must clearly explain that most children will test negative and provide appropriate counselling messages. Even with a sensitivity of over 90% in this high-prevalence population it has a low positive predictive value (PPV) which would be lower still when used in low-prevalence settings
Poor use of the algorithm may be reflective of an overall poor use of IMCI due to: heavy workloads, lack of time for consultation, absence of clinical supervision and support
Poor application of prevention of mother-to-child transmission (PMTCT) programmes. Even though many mothers reported testing positive, few HIV-exposed children had been tested and most clinics did not test children under five. 73% of mothers had been tested for HIV, of whom 24% (221) tested positive. Of the 221 HIV-exposed children, only 35% (78) had been tested for HIV within routine services.The authors recommend the strengthening of PMTCT and linkage with IMCI as well as improved access to HIV PCR for exposed children. This may reduce the need for the algorithm to identify symptomatic HIV. However, it will remain important for children whose mothers do not disclose their status or become infected during pregnancy and breastfeeding, and in settings where virological testing is not available. The authors note that these findings show that undiagnosed HIV infection is common in primary healthcare clinics among under-five year olds and most have advanced disease. Current recommendations suggest that antiretroviral treatment is begun in children under one year of age as soon as HIV status is confirmed. They also note that their findings support the IMCI recommendation to check all children for possible HIV infection. The authors highlight the study’s strengths. The IMCI experts were all highly experienced and provided a “reliable gold standard”; observation of large numbers of children and health workers made it possible to describe performance using the health worker as the unit of analysis. Health workers had no notice of the observation and observation of large numbers over several days reduced bias. The authors note several limitations. The observer’s presence may have influenced performance and led to bias. For example, health workers may have worked out what to do during the observation. Evaluation of individual ability to identify specific signs did not take place to avoid interference during the consultation. Evaluation of the sensitivity of HIV rapid tests in children under 18 months of age remains incomplete. They did not get CD4 results for all HIV-infected children. The authors conclude that IMCI and the correct use of the current guidelines can identify HIV-infected and exposed children and provide increased and earlier access to care in South Africa to reduce under-five mortality. However, poor use of the guidelines limits its potential. The authors suggest further study to understand poor health worker performance “to provide evidence-based interventions to address poor IMCI implementation”.