Poor results using non-medical HIV counsellors to screen for treatment eligibility in Malawi

There may be limits to which medical tasks can be shifted to non-medical staff in resource-limited settings with only limited training and supervision, according to a report from The Lighthouse Trust in Malawi presented at the XVII International AIDS Conference last month.

The report described an attempt to train HIV counsellors to use a symptom checklist to identify patients who could be eligible for antiretroviral therapy (ART) and promptly refer them to ART clinics. But only a fraction of people who were pre-screened as eligible for treatment actually complied with the referral and were then found to be eligible. An analysis of those who followed through on the referral found that counsellors tended to over-refer those with symptoms such as chronic fever and significant weight loss while missing other important symptoms.

However, it is unclear whether the failure was because symptom screening was too complex a task to ask of non-medical personnel, or whether the checklist, the limited training and supervision, or some other factor was to blame.

The use of task shifting in Malawi

Task shifting involves the delegation of healthcare tasks from more highly trained individuals to those with less training, and it has been increasingly employed as a way to help address the shortage of highly trained staff in many resource-limited settings. There is good evidence that clinical officers and nurses can perform many of the tasks done by doctors, and that community health workers can in turn take on many of the tasks usually done by nurses, such as HIV testing and counselling.

Malawi, one of the poorest countries in the world, is experiencing one of the most severe shortages of trained medical staff and so has been one of the pioneers in task shifting. For instance, in 2007, around 625,000 HIV tests were performed, 95% by non-medical counsellors, with 169,000 positive results. In addition, clinical officers and nurses in Malawi can prescribe HIV treatment and run ART clinics, and as a result, the country has been able to launch one of the world’s largest ART programmes. By the end of 2007, there were about 97,000 patients on HIV treatment in 118 ART clinics.

ART eligibility in Malawi is based upon clinical symptom staging, because there is virtually no access to viral load testing, and very limited access to CD4 cell monitoring. This means that people with HIV need clinical evaluations and examinations before they can be put onto treatment — a time-consuming process.

However, “the ART clinics were operating at the very limit of staffing capacity and due to the shortage of healthcare workers, the government of Malawi made a deliberate decision to not have pre-ART clinics for people with HIV,” said Hannock Tweya, who works with the Lighthouse Trust in Lilongwe, and who presented the report.

Unfortunately, this means that people may not come to a clinic until they are quite ill, and some may never make it.

Why not pre-screen?

In an effort to reduce the workload at the ART clinic, the team at Lighthouse Trust decided to see if it would be possible to get counsellors at the HIV counselling and testing sites to screen people testing HIV-positive for more advanced disease and to refer only those people to the ART clinics.

The HIV counsellors are people without a medical background who only undergo a three-week training in counselling and testing.

So the researchers designed a symptom checklist for non-medical HIV testing counsellors based on what their data suggested were the most common symptoms for determining clinical ART eligibility. These included severe weight loss/wasting syndrome, chronic fever, active TB, chronic diarrhoea and oral thrush. They also added Kaposi’s sarcoma because they thought it would be easy for the counsellors to identify.

Then they asked the counsellors to fill this form out for every HIV-positive client and to refer anyone who had at least one positive symptom or other non-listed complaint (which could be entered onto the form). Testing is anonymous in Malawi, so there was no way to follow up people who did not complete the referral, but for those who did, the clinic was able to match them to a symptom checklist by the referral slip ID.

Then Lighthouse compared the assessment done by the counsellor to that of the clinician at the ART clinic, looking at the overall positive predictive value of the checklist for ART eligibility and the agreement for individual symptoms. They also assessed patient compliance with referrals.

Results

Between June 2007 and March 2008, 7183 people tested, 38% of whom were HIV-positive. 92% of these were reported to have been referred on the basis of the checklists; 84% of these chose to be referred to the Lighthouse clinic.

But only 47% followed through on the referral (most within a month, but a small percentage had come in later). Of those coming in within 30 days, more were adults 25 years old and older (adj. OR = 1.69 [1.28 – 2.24]) and those in whom Kaposi's sarcoma was suspected (adj. OR = 1.57 [1.12 – 2.18]). But people with TB treatment history were 70% less likely to present to the clinics (adj. OR = 0.30 [0.24 – 0.37]) — this tendency has previously reported, and may have to do with people only wanting to deal with having one disease at a time.

Of those who acted on the referral, 79% qualified for ART. Of those who were identified with one symptom on the checklist, 52% had WHO stage 3 or 4 HIV disease and 26% had WHO stage 2 disease and a CD4 cell count below 250 cells/mm³. Of those with other complaints, 29% had WHO stage 3 or 4 HIV disease and 31% had WHO stage 2 disease and a CD4 count below 250 cells/mm³. Again, CD4 testing is not widely available in Malawi, though Lighthouse has access to it, and this observation that over a quarter of patients qualified on the basis of CD4 alone is an important one.

As far as the predictive value of the specific symptoms, fever and weight loss were the most commonly identified symptoms on the counsellor checklist but clinicians rarely agreed with these assessments (sensitivity over 80%, specificity 37-42%). A history of TB treatment was less subjective, and thus it is hardly surprising that agreement was moderately good on this symptom. But the counsellors did not ascertain the presence of oral thrush or Kaposi's sarcoma.

However, this may not be surprising given that the counsellors are not trained medical personnel, and were only administering a questionnaire —not performing an examination.

Conclusions

“Almost all clients who came for HIV testing had clinical problems,” said Mr Tweya, “and there was only limited accuracy of assessment by counsellors using the checklist.”

They also concluded that the clinic should look more closely at improving compliance with referral, which might include more client education at the time of HIV testing, specific education for TB patients, and reducing the long waiting times at the clinic.

"Our recommendation is to increase, clinical and lab capacity for pre-ART management,” said Mr Tweya, “but looking at the situation in Malawi, this can only be done within the context of a long-term plan.”

Implications

Reports of this study were circulating earlier in the year, at the HIV Implementers’ meeting in Kampala, where a doctor from Malawi mentioned that counsellors could not perform this task.

However, it is not clear whether this is the right take-home message. HIV counsellors in other settings have been successfully used to perform TB symptom screening — but in most of those cases, programmes are happy if 5% of the people referred actually have TB, so the researchers’ expectations might have been rather high in Malawi.

A problem with this approach is that when people come in for testing, many have symptoms that are worrying them — which they report. These symptoms may not always be chronic, particularly in people who have recently seroconverted. So this may be one reason why there was poor agreement between the checklist and the clinician assessment. Also, how questions are asked is very important on a checklist such as this — a simple `yes` or `no` may not always result in accurate information.

Another conclusion is that thorough clinical evaluation for people with HIV is just one corner that cannot be cut.

Finally, there was no discussion about the training that the counsellors received to fill out the checklist, or what sort of supervision they were provided with during the course of the study. Ongoing supervision and support is recommended for the success of task shifting, and it is not clear how much was given to these counsellors.