Surveillance of Antituberculosis Drug Resistance Among Children From the Western Cape Province of South Africa—An Upward Trend

Abstract

Objectives. We assessed the prevalence of antituberculosis drug resistance among children with tuberculosis (TB) in the Western Cape Province of South Africa.

Methods. Drug susceptibility testing for isoniazid and rifampin was prospectively done on all children with culture-confirmed TB at Tygerberg Children's Hospital, Cape Town, from March 2005 through February 2007. Survey results were compared with results from 2 previous surveys.

Results. We found 291 children had culture-confirmed TB. Resistance to isoniazid or rifampin increased from 21 of 306 (6.9%) to 41 of 319 (12.9%) and 43 of 285 (15.1%) in the first to third surveys (P = .005) and multidrug resistance from 7 of 306 (2.3%) to 18 of 319 (5.6%) and 19 of 285 (6.7%; P = .033). Although previously treated children had significantly more drug resistance than did new TB cases (19 of 66 [28.8%] vs 24 of 225 [10.7%]; odds ratio = 3.39; 95% confidence interval = 1.62, 7.05), evidence suggests transmission rather than acquisition of resistance. HIV infection was not significantly associated with drug resistance.

Conclusions. Results indicate a high and rising prevalence of anti-TB drug resistance among children in the Western Cape, which suggests ongoing transmission of drug-resistant strains within the community. Improved control of TB in adults, including early identification and treatment of drug-resistant cases, is necessary to reduce transmission to children.

Drug-resistant tuberculosis (TB), and more so multidrug-resistant TB (MDR-TB) and extensively drug-resistant TB, pose a major threat to TB control programs.¹ Recent reports of extensively drug-resistant TB and associated mortality emphasize the importance of continued surveillance and the need for measures to prevent development and spread of drug-resistant TB.^2,3

In most developing countries, the diagnosis of adult TB is by sputum smear microscopy. Culture and drug susceptibility testing are not routinely done,^4,5 which precludes ongoing surveillance of drug-resistance patterns within the community. A further challenge is the difficulty of accurate categorization of drug resistance. Recently, the previous classification of primary and acquired drug resistance has been replaced by more pragmatic definitions: new drug resistance (i.e., resistance in cultures from patients who received no previous TB treatment, or treatment for less than 1 month) and previously treated drug resistance (i.e., resistance in cultures from patients previously treated for 1 month or more).⁴ Nonetheless, uncertainty remains regarding classification of drug resistance.

From a TB control perspective, it is imperative to distinguish between primary (i.e., transmitted) drug resistance and acquired (i.e., developed) drug resistance. New drug resistance implies poor control over transmission of disease, whereas development of drug resistance in a patient usually implies poor management of disease. Children with culture-confirmed TB provide an accurate measure of transmitted drug resistance because even children who have received previous treatment for TB tend to have transmitted rather than acquired drug resistance.¹ This is mainly because of the paucibacillary nature of childhood TB, implying a low risk of developing random drug resistance on treatment. Therefore, surveillance of drug resistance patterns among children with culture-confirmed TB provides an accurate measure of transmitted drug resistance within a community.⁶

The prevalence of drug-resistant TB provides an indication of how well TB-control programs are functioning. In particular, trends in new (transmitted) drug resistance, such as is most often found among children, reflect the ability of TB-control programs to limit the spread of resistant organisms within communities and can serve as a sentinel surveillance system.^1,7 We report on trends in the prevalence of drug resistance among children with culture-confirmed TB during 3 surveillance periods at a major referral hospital in South Africa.

METHODS

This study was conducted in the Western Cape Province of South Africa, which reported a TB incidence of 464 per 100 000 in 1998, 988 per 100 000 in 2004, and 1031 per 100 000 in 2006.⁸ The HIV prevalence among women attending public antenatal care facilities in the Western Cape increased from 5.2% (95% confidence interval [CI] = 3.2%, 7.2%) in 1998 to 15.4% (95% CI = 12.5%, 18.2%) in 2004 and to 15.1% (95% CI = 11.6%, 18.7%) in 2006.^9,10

Study Population

The surveys were conducted prospectively among all children younger than 13 years routinely diagnosed with culture-confirmed TB at Tygerberg Children's Hospital. The most recent survey was conducted from March 2005 through February 2007. Previous surveys were conducted from August 1994 through April 1998 and from March 2003 through February 2005.^1,11 The second survey also included a group of patients recruited at 5 primary health care clinics; data from this community-based component were compared with the hospital-based surveillance data to investigate differences between drug resistance patterns in community-based and hospital-based surveys.¹

Drug Susceptibility Testing

Mycobacterium tuberculosis isolates were collected prospectively at the local National Health Laboratory Service microbiology laboratory for all 3 surveys. A single isolate from each patient was sent to the regional National Health Laboratory Service referral laboratory for drug susceptibility testing. Initial drug susceptibility testing was done for isoniazid and rifampin resistance only, and if resistance to either or both drugs was detected, susceptibility testing for ethambutol was done according to the South African National Tuberculosis Program policy at the time. Middlebrook 7H9 broth base (Mycobacterial Growth Indicator Tubes; Becton Dickinson, Sparks, MD) culture medium was used for primary isolation. The presence of M tuberculosis was confirmed by polymerase chain reaction amplification.¹²

During the most recent survey, 2 different laboratory procedures for drug susceptibility testing were used because routine drug susceptibility testing was introduced during 2006 following the high prevalence of drug-resistant TB recorded during the 2003 to 2005 survey. The different methods yielded very similar results.¹³ Drug susceptibility testing procedures were as follows: (1) method 1: drug susceptibility testing by the indirect proportion method on Middlebrook 7H10 agar medium containing critical concentrations of 0.2 μg/mL of isoniazid, 30 μg/mL of rifampin, and, where indicated, 2 μg/mL of ethambutol; (2) method 2: drug susceptibility testing with the Bactec 460TB system (Becton Dickinson, Sparks, MD), according to international criteria.¹⁴ Isoniazid was tested at a concentration of 0.1 μg/mL, rifampin at 2.0 μg/mL, and ethambutol at 7.5 μg/mL. The susceptibility of a strain was judged by comparing growth of organisms in drug-containing compared with nondrug-containing media; resistance was defined as 1% or more bacterial growth in the drug-containing media. Quality assurance for drug susceptibility results was done locally with every batch and quarterly by the national TB reference laboratory.^1,11

Clinical Data and Treatment of the Recent Survey

The clinical records of all children with a positive M tuberculosis culture were reviewed. All children with drug-resistant TB were recalled. A history was obtained from parents of all children regarding previous TB prophylaxis or treatment and whether the children had a history of close contact with adult pulmonary TB.

HIV testing was done at the discretion of the attending physician, following written informed consent from the parent or legal guardian. HIV infection was confirmed by 2 different enzyme-linked immunosorbent assays and by polymerase chain reaction in children younger than 18 months.

Children with drug-susceptible TB were treated according to the South African National Tuberculosis Program guidelines with isoniazid, rifampin, and pyrazinamide.⁵ Children with drug-resistant TB were treated according to their specific drug susceptibility pattern for 12 to 18 months after the first negative culture result; all children were followed up by H. Simon Schaaf at the pediatric drug-resistant TB clinic at Tygerberg Children's Hospital.

We used Epi Info version 6.04 (Centers for Disease Control and Prevention, Atlanta, GA), for statistical analysis. Categorical data comparing proportion of children with drug resistance, HIV infection, and previous TB treatment among the 3 surveys and drug resistance between groups who were HIV infected and groups who were not HIV infected in the last 2 surveys combined were analyzed with the χ² test. The Fisher exact test was applied when appropriate.

RESULTS

During the most recent survey, 291 children (154 boys; 52.9%) were given the diagnosis of culture-confirmed TB. The median age was 33 months (range = 7 days–156 months). Demographic and clinical data for all 3 surveys are summarized in Table 1. Drug susceptibility test results for the 3 surveys are compared in Table 2. Any drug resistance increased from 21 of 306 (6.9%) children to 41 of 319 (12.9%) and to 43 of 285 (15.1%) in the first, second, and third surveys, respectively.

TABLE 1.

Demographic and Clinical Data From Children From the Western Cape Province of South Africa With Culture-Confirmed Tuberculosis (TB) From 3 Surveys: August 1994–April 1998, March 2003–February 2005, and March 2005–February 2007

	1994–1998 (n = 338)	2003–2005 (n = 323)	2005–2007 (n = 291)	Pa
Median age, y (range)	2.6 (0.1–13.0)	2.5 (0.04–12.9)	2.8 (0.02–13.0)
Boys, no. (%)	193 (57.1)	173 (53.6)	154 (52.9)	.52
Previous TB treatment > 1 mo, no (%)	32 (9.5)	57 (17.6)	66 (22.7)	<.001
History of TB contact, no. (%)	138 (40.8)	174 (53.9)	159 (54.6)	<.001
HIV test done,b no. (%)	166 (49.1)	243 (75.2)	184 (63.2)	<.001
HIV infected,c no (%)	13 of 166 (7.8)	64 of 243 (26.3)	49 of 184 (26.6)	<.001
Type of TB, no (%)
Pulmonary	NA	281 (87.0)	247 (84.9)	.52
Extrapulmonaryd	NA	169 (52.3)	169 (58.1)	.18
Both	NA	127 (39.3)	125 (43.0)	.41

Note. NA = not available.

^aP values compare differences among all 3 groups.

^bSignificant decline in HIV tests done from second to third survey (odds ratio = 1.77; 95% confidence interval = 1.23, 2.54).

^cIncrease between the last 2 surveys was not significant.

^dDefined as TB manifestations outside the thorax, including pleural and pericardial effusions.

TABLE 2.

Drug Susceptibility Test Results for the 3 Surveys in the Western Cape Province of South Africa: August 1994–April 1998, March 2003–February 2005, and March 2005–February 2007

Drug Susceptibility Test Results	1994–1998 (n = 338), No. (%)	2003–2005 (n = 323), No. (%)	2005–2007 (n = 291), No. (%)	Pa
Drug susceptibility test available	306 (90.5)	319 (98.8)	285 (97.9)	<.001
Drug susceptibleb	285 (93.1)	278 (87.1)	242 (84.9)	.005
Any resistanceb	21 (6.9)	41 (12.9)	43 (15.1)	.005
Isoniazid monoresistance	14 (4.6)	23 (7.2)	22 (7.7)	.24
Rifampin monoresistance	0	0	2 (0.7)
Multidrug resistancea	7 (2.3)	18 (5.6)	19 (6.7)	.03

^aP values compare differences among all 3 groups.

^bDifference between last 2 surveys was not significant.

In the most recent survey, 29 of 49 (59.2%) children who were HIV infected had previously had more than 1 month of anti-TB treatment compared with 37 of 242 (15.3%) children who were not HIV infected or who had unknown HIV status (odds ratio [OR] = 3.82; 95% confidence interval [CI] = 1.86, 7.86). Drug resistance occurred in a significantly higher proportion of children who had received previous treatment for TB (19 of 66; 28.8%), compared with new TB cases (24 of 225; 10.7%; OR= 3.39; 95% CI = 1.62, 7.05). However, clinical and laboratory evidence suggested transmission rather than acquisition as the mechanism for developing drug resistance in at least 14 of the 19 previously treated drug-resistant children.

Of the 19 children with drug resistance who had received previous treatment for TB, 6 had isoniazid monoresistance, 2 had rifampin monoresistance, and 11 had MDR-TB. Of these, 5 MDR-TB contacts who received isoniazid or isoniazid and rifampin chemoprophylaxis and were adherent, and all had MDR-TB; 3 parents (2 mothers, 1 father) had confirmed MDR-TB, and their children were not adherent to first-line treatment; and 1 child with MDR-TB was confirmed to be reinfected with a drug-resistant strain by DNA fingerprinting. Five children—2 with MDR-TB, 2 with rifampin monoresistance, and 1 with isoniazid monoresistance (2 of whose mothers died with no drug susceptibility testing done)—had progressive disease despite adherence to first-line treatment. In 3 of the remaining children, adherent TB treatment was completed more than 1 year before; all were isoniazid-resistant only, and 2 had new contacts (1 died) with unknown drug susceptibility testing results. The last 2 children were treatment defaulters who received irregular treatment, 1 of whom was HIV infected and had a CD4 cell count of less than 1%. Both had isoniazid monoresistance, most likely acquired because of poor adherence. In total, at least 38 of 43 (88.4%) of all drug-resistant children in this study and all participants with MDR-TB most likely had transmitted drug resistance.

Although 159 (54.6%) children had a known TB source case, drug susceptibility testing results of the source case were available in only 24 (15.1%). Of the 24 children with drug susceptibility testing results known in their source cases, 17 had source cases with MDR-TB (3 of whom also had drug-susceptible source cases), and 7 had drug-susceptible source cases. Of the 17 contacts with MDR-TB, all 3 children who also reported contact with a drug-susceptible source case had drug-susceptible TB; 8 of the children in contact with MDR-TB cases had MDR-TB, 3 had isoniazid-monoresistant TB, and 3 had drug-susceptible TB. A 57.1% agreement (8 of 14) was found between having MDR-TB disease and having a reported contact with only an MDR source case.

Fourteen (4.8%) children died while in the hospital; 4 (3 MDR) had drug-resistant TB (3 were HIV infected), 4 had stage 3 TB meningitis (1 MDR—also included in the 4 drug-resistant deaths), 4 were severely malnourished (2 were HIV infected, both were marasmic, and 2 had kwashiorkor), 1 had congenital TB, 1 had miliary TB, and 1 had systemic lupus erythematosus.

HIV infection increased from 13 of 166 (7.8%) to 64 of 243 (26.3%) and to 49 of 184 (26.6%) of the children tested for HIV from the first, second, and third surveys, respectively. During the last 2 surveys, 113 of 614 (18.4%) children received the diagnosis of HIV infection, 314 (51.1%) were not HIV infected, and HIV status was not known in 187 (30.5%). Of the children who were HIV infected, 21 (18.6%) had some drug resistance compared with 47 (15.0%) of the children who were not HIV infected and 16 (8.6%) of the group with HIV status not known. No statistically significant difference in drug resistance was seen between the group that was HIV infected and the group that was not infected (OR = 0.76; 95% CI = 0.42, 1.40) or between the group that was HIV infected and a combination of the group that was HIV infected and the group with HIV status not known (OR = 0.63; 95% CI = 0.36, 1.12).

DISCUSSION

To our knowledge, this was the first series of TB drug susceptibility testing surveys of isolates from children reported from a developing country setting. Steiner et al.^15–17 conducted a series of hospital-based surveys in New York from 1961 to 1991, showing an increase in both isoniazid and multidrug resistance. In this study, we could report with reasonable certainty that approximately 88% of all drug resistance and almost all MDR-TB in children was a result of transmitted infection rather than acquired resistance, despite the high percentage of retreatment cases.¹ Children with culture-confirmed TB, whether previously treated or not, therefore provide a good indication of currently circulating strains in a community.¹⁸

Reports of drug resistance in children are mainly hospital based,^16,19–23 for which bacteriological confirmation is more readily achieved. Although hospital-based studies may introduce bias because children with more-severe illness, HIV infection, and previous TB are more likely to be referred to the hospital, a previous survey at our hospital found no difference in drug susceptibility rates between children recruited from community-based clinics and children recruited from the local referral hospital.¹

We note with concern that drug resistance among children with culture-confirmed TB at this hospital follows an upward trend, although the increase between the last 2 surveys was not statistically significant. Increased awareness of drug-resistant TB and the fact that our hospital has a dedicated clinic for children with drug-resistant disease may have introduced some bias in detection of patients with drug resistance. However, this is unlikely to have influenced the observed upward trend because this specialist clinic was operational throughout all 3 study periods.

TB drug susceptibility surveys conducted among adults in the Western Cape during 1994 and 2001 have shown similar results (4% new isoniazid resistance and 1% new MDR patients) and did not indicate an increase in drug resistance among adults.^7,24 However, the number of adults with TB and MDR-TB has increased dramatically since the last survey was conducted in 2001. Because routine drug susceptibility testing is now done on all children who receive a diagnosis of culture-confirmed TB at the 2 major pediatric referral hospitals in the Western Cape Province, pediatric data generated from surveys such as ours are key sources of TB surveillance data, especially because routine adult surveillance is not readily available.¹³

The increase in HIV infection in the children with TB in these surveys followed a trend similar to that of the HIV prevalence among women attending public antenatal care facilities in the Western Cape, which increased from 5.2% in 1998 to 15.4% in 2004 and to 15.1% in 2006.^9,10 Despite a high prevalence of HIV infection in hospitalized children with TB, HIV testing decreased significantly from the second to the third survey. Because of the difficulty in diagnosing TB in children who are HIV infected, the World Health Organization has recommended that all children with suspected TB living in a high HIV prevalence area or children who are at high risk for HIV infection should have an HIV test done.²⁵ The Western Cape Province remains a high HIV prevalence area; in such settings, all children suspected of having TB should be tested for HIV.

Drug susceptibility testing results of the presumed source case usually provide a good indication of the child contact's drug susceptibility pattern, with more than 80% of children having the same strain of MDR-TB as their source case; therefore, children in contact with adults with drug-resistant TB should be treated according to the source case's drug susceptibility testing result.^26–28 As found in the current study, in areas with an extremely high incidence of TB, exposure to more than 1 source case, some with drug-resistant and others with drug-susceptible TB, is not uncommon. However, the high failure rate (5 of 5 cases) of isoniazid or isoniazid and rifampin chemoprophylaxis and first-line therapy (3 of 3 cases) in children with known MDR-TB household exposure indicates the importance of treating child contacts of known MDR-TB source cases according to the drug susceptibility testing result of the source case until (and if) an isolate of the child is obtained for drug susceptibility testing.

It is of great concern that despite relatively high rates of drug-resistant disease, drug susceptibility testing results are rarely available from adults with TB in this area. According to current national guidelines, routine drug susceptibility testing is indicated for all retreatment cases. This policy is often not followed, because of a lack of human capacity within the health care system, a lack of awareness of the importance of rapid identification of drug resistance, and the difficulty in obtaining further sputum specimens for culture and drug susceptibility testing after smear microscopy has been done. In addition, our data suggest a high prevalence of more than 5% transmitted (primary) multidrug resistance, which may indicate a need for routine drug susceptibility testing of all adults with TB.

Our results indicate a high and rising prevalence of anti–TB drug resistance among children in the Western Cape Province of South Africa. This suggests ongoing transmission of drug-resistant strains within the community and is cause for serious concern. To prevent further increases in drug-resistant TB among children, improved case finding among adults; more-rapid diagnosis of drug-resistant TB, even in new adult TB cases; and initiation of early appropriate treatment regimens in adults are necessary to shorten the time children are exposed to infectious drug-resistant patients. An urgent need also exists for the evaluation of appropriate chemoprophylactic regimens for child contacts, especially those of patients with MDR-TB.