Multi-drug resistant TB: breakdown in counselling often to blame

Although the incidence is still fairly low, there is an increasing rate of resistance to TB drugs in patients in Botswana, and much of this may be the result of premature treatment discontinuation due to inadequate counselling and a widespread belief in Botswana that TB is not fatal, according to two studies presented at the 1st National HIV and AIDS (NHASORC) last month in Gaborone, Botswana.

In addition to the very high prevalence of HIV infection, Botswana is burdened with a very high caseload of TB. In 2001, the yearly incidence of TB was approximately 620 cases per 100,000 persons according to studies presented Like the HIV epidemic, this increasing incidence is a relatively recent phenomenon. After the establishment of a national TB programme in 1975, Botswana had seen a steady decline in the TB incidence. But since 1989, the TB case rate in Botswana has increased simultaneously with rising rates of HIV.

Several studies on TB and its relationship to the HIV epidemic were presented at the Gaborone meeting. The first two evaluated possible causes of treatment failure: drugs resistance and premature treatment discontinuation. Two other reports at the meeting investigated the epidemic in two “problem” populations: prisoners and refugees (see TB in hard-to-reach populations), while another reported disappointing results reviewing different rapid diagnostic test for TB (see TB diagnostics in Botswana).

Dr. Paul Ngirubiu reported the results from the most recent TB drug resistance survey, conducted from March to November 2002. Previous surveys performed in Botswana in 1995-to 6 and 1999 showed very low rates of drug resistance. However, as some studies have found an association between HIV and TB drug resistance (though others have not), there are concerns that HIV could increase TB drug resistance.

The survey measured the rates of resistance to first-line TB drugs and at the same time conducted anonymous HIV surveillance to determine whether it was linked to increased drug resistance. The survey also assessed the prevalence of non-tuberculous mycobacteria (NTM).

Specimens were collected from labs all over the country, from 2425 patients who developed TB during the survey period. The median age was 34 yrs and 1333 of the patients (55%) were male. Overall, 1455 (60%) of the sputum specimens were HIV-positive by rapid testing. Culture results are available on 61% of the total isolates: 83% of these were mycobacterium tuberculosis while 2% were positive for nontuberculous mycobacteria (results are still pending on the other 15%).

TB cases occurred in 1124 new patients in the survey — 10.3% had some drug resistance. This represented a statistically significant increase from previous survey findings (3.7% isolates in 1995-6, and 6.3% in 99).

Isoniazid and rifampicin are the most important TB medications used in Botswana. The survey found that 4.4% of the new patients had some isoniazid resistance, and 1.9% had some resistance to rifampicin. Resistance to both drugs was observed in 0.8% of the new patients, which represented an increase from 0.2% in previous surveys (this finding did not reach statistical significance: p=0.17).

Among 100 previously treated patients, resistance had increased since the previous surveys, from 14% and 22.8% to 24%. This finding did not quite reach statistical significance (p=0.08), but the number of patients involved was small. 15% and 13% of the resistant-pretreated patients were resistant to isoniazid and rifampicin, respectively.

MDR-TB is also on the rise among previously treated patients, up from 5.8 and 9% in the two previous surveys to 11%. Although clearly trending upwards, this finding did not reach statistical significance (p=0.16)

In a very preliminary analysis, HIV status was not associated with a higher risk of developing drug resistance (RR=0.60) (95% CI 0.3, 1.7). However, the study was poorly powered to answer the question (out of the 2425 patients enrolled, they have received only 76 follow-up specimens, only 61 of these isolates had drug resistance data thus far that could be compared to HIV status. Furthermore, it was pointed out by members of the audience that HIV’s effect on resistance could be quite complex. For example, the infectiousness of people with drug resistant TB can vary according to stage of HIV disease. Also, the very burden of the HIV epidemic could so tax the healthcare infrastructure that under-treating TB becomes more common, leading to TB drug resistance in more individuals whether they are co infected with HIV or not.

The second presentation on TB addressed this issue of under-treatment. Although Botswana has a strong direct observed therapy program using a short (DOTS) course strategy for TB control, according to Dr. B Chengeta, a study in 1998 found that at least 10% of TB patients discontinue prematurely. Treatment interruption contributes to poor patient outcomes, chronic disease or relapse, multidrug-resistance and the need for more costly second-line treatment.

Considering the high case rate, treatment discontinuation could pose a significant public health issue. In order to better understand why people quit treatment, B. Chengata and colleagues designed a case-control study to investigate what factors were associated with treatment interruption. They conducted a survey of patients selected from Botswana’s electronic TB registry (ETR), a computerised TB database developed to assess patient management and programme performance. (According to another report at the conference, the ETR was based on WHO recommendations for quarterly surveillance of TB and has recently been upgraded from DOS to WINDOWS format; a version of it is now in use by most of the countries in the southern African region).

The researchers wanted to include 399 adult patients diagnosed with TB in 2000, to compare at least 133 treatment completers, classified as patients who had missed fewer than 21 days of therapy, with 266 non-completers.

The researchers found, that although the ETR can be a useful research tool, inadequate follow-up can lead to a lot of misclassification. When the classifications were confirmed (by record review and/or interview) it was found that a large number of the patients had been misclassified in the ETR. For example, the majority of those misclassified as non-completers had actually died on (or very shortly after discontinuing) treatment.

This left only 97 confirmed non-completers with 130 confirmed completers (plus those who had been originally misclassified). A study nurse then attempted to interview as many of these as possible using a structured questionnaire administered in the patient’s native language. Unfortunately, many patients died before they could be reached. In the end, only 63 non-completers could be interviewed.

Nevertheless, the questionnaire produced some interesting findings. For example, 34% of the patients reported receiving no counselling about TB and 23% reported that the delivery of directly observed therapy was inconsistent. Of the non-completers treatment, 44% interrupted during the initial phase of treatment. 58% reported side effects were the major reason for interruption.

With a median age of 33 (range 18-84), little over half (53%) of the participants were male, but they were far more likely to discontinue treatment than women (OR 2.4, confidence interval 1.3-4.7). Those who were treatment non-completers were also more likely to drink alcohol (OR 3.9, CI 1.7-10), and to have what they described as a chaotic life (OR 2.1, CI 1.1-4.0).

When compared to treatment completers, non-completers were more likely to have had a delayed visit to the clinic (134 days vs. 78 days, p<0.05) and to have received no TB counselling (77% vs. 23%, p<0.05). Traditional healer consultation was common and in both groups, patients often did not believe TB could cause death.

In conclusion, Dr. Chengeta recommended that to better manage TB in Botswana, further efforts are needed “to ensure that there is adequate counselling at TB diagnosis and to investigate ways to increase quality of DOT.” She also believes that the cause of death after treatment needs to be routinely investigated and reported — for example, is a death due to TB relapse or death from other infections? This in turn would help make sure that cases are not misclassified in the ETR and that the database can be a reliable tool for programme assessment.