Skip Navigation

Molecular Epidemiology of Tuberculosis in the Netherlands: A Nationwide Study from 1993 through 1997

  1. Dick van Soolingen1,
  2. Martien W. Borgdorff3,
  3. Petra E. W. de Haas1,
  4. Maruschka M. G. G. Sebek3,
  5. Jaap Veen3,
  6. Mirjam Dessens1,
  7. Kristin Kremer1 and
  8. Jan D. A. van Embden2
  1. Diagnostic Laboratory for Infectious Diseases and Perinatal Screening
  2. Research Laboratory for Infectious Diseases, National Institute of Public Health and the Environment, Bilthoven
  3. Royal Netherlands Tuberculosis Association, The Hague, Netherlands
  1. Reprints or correspondence: Dr. Dick van Soolingen, Diagnostic Laboratory for Infectious Diseases and Perinatal Screening, National Institute for Public Health and the Environment, P.O. Box 1, 3720 BA Bilthoven, Netherlands (D.van.Soolingen{at}rivm.nl)
 
Next Section

Abstract

To disclose risk factors for active tuberculosis transmission in the Netherlands, restriction fragment length polymorphism (RFLP) patterns of 78% of the Mycobacterium tuberculosis isolates, from the period 1993–1997, were analyzed. Of the respective 4266 cases, 46% were found in clusters of isolates with identical RFLPs, and 35% were attributed to active transmission. The clustering percentage increased strongly with the number of isolates; taking this into account, fewer cases were clustered than has been reported in other studies. Contact investigations in the five largest clusters of 23–47 patients suggested epidemiological linkage between cases. Of patients identified through contact tracing, 91% were clustered. Demographic risk factors for active transmission of tuberculosis included male sex, urban residence, Dutch and Surinamese nationality, and long-term residence in the Netherlands. Human immunodeficiency virus infection was not an independent risk factor for active transmission. Isoniazid-resistant strains were relatively less frequently clustered, suggesting that these generated fewer secondary cases.

Molecular strain typing of Mycobacterium tuberculosis by restriction fragment length polymorphism (RFLP) has been used to study the epidemiology of tuberculosis. Disclosure of patients harboring M. tuberculosis strains with identical RFLP patterns suggests an epidemiological link between such “clustered” cases, provided that sufficient genetic diversity among M. tuberculosis strains exists in the respective population. RFLP typing has been used for various epidemiological purposes, such as outbreak investigations in health care settings and communities [1–3], to determine risk factors for active transmission of tuberculosis [4–6], to estimate the proportion of epidemiological links not identified by conventional contact tracing [4, 6], to identify misdiagnosis due to laboratory cross-contaminations [7–9], and to quantify tuberculosis transmission between subpopulations [10]. Population-based studies in which all M. tuberculosis isolates during a given time period are subjected to RFLP typing have been shown to be particularly useful in disclosing unsuspected cases of transmission and quantifying ongoing transmission in a population [4, 10]. As a result of an intensified tuberculosis surveillance in the Netherlands, we typed all M. tuberculosis isolates from patients who developed tuberculosis since 1993. In this report we describe the analysis of the molecular epidemiological findings during the first 5 years of the application of systematic RFLP typing.

Traditionally, the incidence of tuberculosis due to active transmission has been estimated indirectly from notification data [11, 12]. These studies suggest that tuberculosis in countries with low prevalence is mainly attributable to reactivation of remote infections acquired in the past, in particular in patients in older age groups.

Population-based studies using RFLP typing of M. tuberculosis have indicated an unexpectedly high degree of ongoing transmission in regions with low incidence [4, 13]. Furthermore, these studies showed that the proportion of nonclustered tuberculosis cases increases with age, which is consistent with a higher frequency of reactivation of remote infections in the older age groups [4–6, 13, 14]. In this study we analyzed to what extent the age dependency of clustering is due to a decreasing incidence of active transmission with increasing age or to an increasing incidence of reactivation.

The proportion of clustered cases has been used to determine the contribution of active transmission to the total incidence of tuberculosis [4, 5, 13–15]. However, the number of patients included, the age structure of the human population, and the study period differed greatly in these population-based studies. As these factors are expected to greatly influence clustering [16], the degree of ongoing transmission as estimated from the percentage of patients clustered is difficult to compare in the various studies. The large number of patients analyzed in the current long-term study enabled quantification of the influence of such factors.

The assumption that clustering is due to active transmission has been only partly validated. In outbreak investigations, the epidemiological links between cases sharing identical RFLP patterns are clear [1–3]. In citywide studies, epidemiological links between cases have been found, such as shared-risk environments like homeless shelters or shared-risk behavior like alcohol or drug abuse [4–6, 17–21]. In a rural area in Arkansas, epidemiological links in contact tracing were not found among 58% of the clustered cases, despite extensive contact investigations [22]. In this study we attempt to clarify to what extent clustering represents active transmission in a country with low prevalence.

In Europe in general, and in the Netherlands in particular, an increasing proportion of tuberculosis cases occurs among immigrants [23, 24]. In a recent study, using a new analytical approach, we estimated that 17% of the tuberculosis cases among the Dutch were attributable to active transmission from non-Dutch source cases [10]. The transmission attributable to Dutch sources was about equal: 15% [10]. However, this study did not describe in detail which Dutch subgroups interacted with particular immigrant populations. Therefore, we now investigated the characteristics of these “bridging populations.”

Drug resistance of tuberculosis bacilli is becoming a major global problem, leading to a reduced effectiveness of currently available treatment regimens [25]. Inadequate prescription, noncompliance, and insufficient drug supply all contribute to increasing levels of drug resistance [25]. Drug-resistant strains have been found to spread rapidly through nosocomial transmission among human immunodeficiency virus (HIV)—infected individuals [1, 18]. Studies in San Francisco and New York City showed that drug-resistant M. tuberculosis was more prevalent among clustered than among nonclustered isolates. This is in contrast to the situation in the Netherlands, where isoniazid-resistant strains were found to be transmitted less frequently than isoniazid-susceptible strains [26]. However, from this study, it was unclear to what extent this difference was due to confounding factors such as differences in patient populations. In the Netherlands, isoniazid resistance is particularly prevalent among recently immigrated residents. In a recent study based on RFLP typing results and information on previous treatment, transmission of drug-resistant strains in the Netherlands was found to be uncommon [27]. The large number of patients included in this study allowed us to investigate the relationship between drug resistance and clustering in more detail.

Previous SectionNext Section

Methods

Study population

In the period from January 1993 to December 1997, tuberculosis was diagnosed in 8183 patients in the Netherlands. In total, 5676 cases were culture-positive (69%). We excluded 330 cases because of incomplete laboratory or patient information and 224 cases from whom M. bovis or M. bovis bacilli Calmette-Guérin was isolated. Thus, 5122 patients were available, and M. tuberculosis isolates from these cases were subjected to RFLP typing.

Demographic and clinical data were obtained from the National Tuberculosis Register (NTR), held by the Royal Netherlands Tuberculosis Association, which receives data from municipal health services. These included data about demography, localization of disease, earlier treatment for tuberculosis, case-finding method, bacteriology, and risk groups. For privacy reasons, the NTR does not contain patient names and addresses. We therefore matched RFLP data with those in the NTR on the basis of the postal code, date of birth, and sex. This procedure resulted in 4357 matching patients (85%). No bias among nonmatched patients was found in sex, age, or proportion of isolates clustered. Matching was less complete for isolates with multidrug resistance (34 [72%] of 47) and unknown drug sensitivity (44 [54%] of 81) than for the others (4279 [86%] of 4994; P < .001).

Of the 4357 patients with matching patient information and RFLP results, 91 had a unique RFLP pattern in the remaining database, although they were clustered with individuals for whom the matching process had failed or with individuals outside the study period. To remove ambiguity, these were excluded from the analysis. In total, 4266 patients were included in the main data analysis. This comprised 78% of the culture-positive cases caused by M. tuberculosis in the study period. The full 5-yr period was used to define clustering. Population data for calculation of incidence rates were obtained from the Netherlands Central Bureau of Statistics.

Contact tracing to examine possible epidemiological linkage of clustered cases was done in the period from December 1994 to December 1996 [28]; 1170 clustered cases were available for this investigation of epidemiological relatedness.

Drug resistance testing

Drug susceptibility of the M. tuberculosis isolates was determined by use of the minimal inhibition concentration method [29]. In the matched data set, information on drug susceptibility was available for 99% of the isolates. Of these, 3.4% showed monoresistance to isoniazid, 4.9% to streptomycin, and 0.2% to rifampin. Combined resistance to isoniazid and streptomycin occurred in 2.8% of the isolates and to streptomycin and rifampin in 0.1% of the strains. Multidrug resistance (MDR), defined as resistance to at least isoniazid and rifampin, was found in 0.8% of the isolates. Fifty percent of these MDR strains were in addition resistant to streptomycin.

DNA fingerprinting

All M. tuberculosis cultures were subjected to the standardized IS6110-based RFLP typing, as described elsewhere [30]. Because the differentiation of M. tuberculosis strains carrying few IS6110 copies is poor [31], all 433 strains (8.5%) harboring ⩽4 IS6110 copies were subjected to subtyping by use of the Polymorphic GC-rich Sequence (PGRS) as a probe [31, 32]. Computer-assisted analysis of RFLP patterns was done by use of the GelCompar software, version 4.1 for Windows (Applied Maths, Kortrijk, Belgium) [33, 34]. The Dice coefficient was used to calculate the similarity coefficients [33, 34], and cluster analysis was performed by use of the unweighted pair group method with the arithmetic averages method. Clusters were defined as groups of patients having isolates with identical RFLP patterns, that is, the same number of IS6110 copies at identical band positions. For isolates with <5 IS6110 bands, an additional requirement for clustering was the same number of PGRS-containing restriction fragments at identical band positions.

Epidemiological evidence on clustering representing active transmission

Epidemiological evidence for the hypothesis that clustering represents active transmission (with the exception of the original source case of a cluster) was sought as follows. First, we examined what proportion of cases identified through traditional contact tracing had RFLP patterns identical to the index case. This information was obtained from regional public health services. Second, we determined what proportion of clustered cases was found to be epidemiologically linked to other cases in the respective cluster in routine contact investigations. Finally, we analyzed the sociod-emographic characteristics of patients in the 5 largest clusters, to assess to what extent identical RFLP patterns might be due to chance occurrence of predominant strains.

Statistical analysis

Incidence rates were calculated using midyear populations to estimate person-years at risk by age, sex, and nationality. The influence of sample size and study period was determined by taking subsamples from the total study population either by random or by study period. Differences between proportions were tested for statistical significance with the χ2 test. Comparisons between clustered and nonclustered cases were expressed as odds ratios (ORs), and adjustment for confounders was done with logistic regression.

Previous SectionNext Section

Results

Degree of clustering and incidence rate of clustered tuberculosis by age and nationality

The study population was composed of 4266 patients who developed tuberculosis during the period 1993–1997. A unique M. tuberculosis RFLP pattern was found among 2295 (54%) of the patients, the “nonclustered” cases. The remaining 1971 (46%) cases shared an identical RFLP pattern with ⩾ 1 other patients, the “clustered” cases. The clustered cases was composed of 479 clusters, each having a distinct RFLP pattern.

The number of patients within a cluster varied from 2 to 47, and 52% of the clustered cases were in small clusters of ⩽5 patients (figure 1). Assuming that each cluster contains one source case [4] and that all other cases in the clusters were due to active transmission, we calculate that 35% of the culture-positive cases were caused by active transmission of tuberculosis in the 5-year period.

Figure 1
View larger version:
Figure 1

Number of patients by cluster size. Total number of clustered patients during 1993–1997 was 1971.

The incidence rate of nonclustered and clustered tuberculosis by age and nationality is presented in figure 2. Among the Dutch, the incidence rate of nonclustered tuberculosis increased steeply with age. In contrast, not much age-dependent variation in the incidence rate was found among the clustered cases (figure 2). The incidence rate of tuberculosis among the non-Dutch was 10–26 times higher than among the Dutch (figure 2). Furthermore, the incidence rate of nonclustered tuberculosis among the non-Dutch decreased strongly with age ⩽55 years and increased sharply from ages >65 years. The incidence rate of clustered tuberculosis among the non-Dutch showed a similar age dependency. However, no increase was found for ages >65 years. Compared with Dutch patients, a smaller proportion of non-Dutch patients was clustered (figure 2). Tuberculosis occurred more frequently among males than among females both in the Dutch and non-Dutch (data not shown), and clustered cases were also more frequent among males (data not shown).

Figure 2
View larger version:
Figure 2

Percentage of clustered cases by age and nationality (A and C, number of cases, n, of each age class is depicted below x-axis) and incidence rate (cases per 100,000 per year) of clustered and nonclustered tuberculosis by age and nationality (B and D).

Sample size and study period as determinants of proportion of clustered cases

To estimate the influence of sample size and sampling time on the apparent degree of clustering, we compared the degree of clustering from incremental subsets of the 4266 tuberculosis cases. As shown in figure 3, there was a strong effect of the sample size of randomly selected samples on the percentage of clustering. This percentage increased from 12% for a subset of 200 samples to 46% for all 4266 patients. Increasing the number of cases by increasing the time window (measuring the time-dependent plus size-dependent variation) showed a higher percentage of clustering than random samples of the same size, starting from 22% in the first quarter of 1993. These data indicate that the effect of increasing the study period on the increasing percentage clustered is largely explained by the increasing number of isolates (figure 3). The additional clustering is due to the tendency of clustered cases to occur within a limited period of time, but this effect dominates only during the first 6 months.

Figure 3
View larger version:
Figure 3

Percentage of clustering of cases found in practice at different time intervals during the study period (solid line) and clustering of cases chosen randomly (dashed line). Number of cases in latter approach for each sample (bulleted line) was chosen to be equal to number of samples in incremental study periods.

Epidemiological evidence on clustering representing active transmission

In the study period, 215 patients were identified through contact tracing, and 195 (91%) of these were found in clusters. In contrast, only 46% (1507/3314 patients) of the cases detected through passive case finding were in clusters (crude OR, 12). We were able to epidemiologically evaluate 157 of the 168 clustered cases identified through contact tracing in the period 1993—1996. For 130 cases (83%), an epidemiologic connection could be identified with at least 1 other patient in the cluster, whereas 16 patients (10%) had a possible link with other patients because they shared the same places of entertainment. No connection with other cases in the cluster could be found in only 11 (7%) of the 157 cases.

In the Netherlands, all clustered tuberculosis cases are reported to the public health services, which in turn back-report the results in contact tracing of these cases. We were therefore able to evaluate the epidemiologic findings of 1170 clustered patients, of whom 523 (45%) had been involved in a routine contact tracing program. Of these 1170 clustered patients, 361 (31%) had a known contact with at least one other case identified in the corresponding cluster, and 162 (14%) had a possible contact. Possible contacts were defined as people sharing the same environment (shop, shelter, etc.) without identifying the other person as a contact. No obvious contact with patients in the corresponding cluster was found in 647 (55%) of the cases. The large majority (87%) of identified contacts were among household members, family, friends, or colleagues, suggesting that underdetection of casual contacts in these routine follow-ups was likely [28].

During the study period, 2774 different RFLP patterns were observed among the 4266 isolates. The probability of identical RFLP patterns occurring by chance would be expected to be small, except for certain commonly occurring RFLP types. To assess this possibility, we examined the 5 largest clusters (clusters A—E) composed of 47, 43, 41, 34, and 23 patients, respectively (figure 4). Cluster A was composed of 47 cases, and the majority of these cases were found to be epidemiologically linked, as described in detail elsewhere [3]. An infectious patient, who probably excreted M. tuberculosis during the 5 months before tuberculosis was diagnosed, initiated this microepidemic. Between 1993 and 1995, contact tracing resulted in disclosure of over 200 infected people and about 50 active tuberculosis cases. The majority of the initial and satellite cases were geographically related to a few cities where the index patient resided [3]. As shown in figure 4, cluster A was composed mainly of Dutch patients, and few of these appeared to be at increased risk, such as homelessness and drug abuse. Of the 84 patients in clusters B and C, 55 (65%) were from Amsterdam, and the characteristics of the patients in Amsterdam have been described [6]. Cluster C was composed of a high proportion of hard drug users, HIV-positive individuals, and homeless people. Cluster D was composed of 34 cases of which 30 (88%) were Somali patients, 3 were from Ethiopia, and 1 was a Dutch physician who had worked in a Somali refugee camp in Kenya. Cluster E, composed of 23 patients, evolved in a relatively short period of 30 months. The first cases related to this cluster were caused by transmission from an asylum seeker to 7 other asylum seekers of different origin. The remaining cases were found among Dutch residents. How the transmission from the asylum seekers to the native population occurred was not completely clear but most likely took place in the soft drug circuit. Although the evidence falls short of proof, the pattern observed among the 5 largest clusters strongly suggests epidemiological linkage rather than the common occurrence of predominant strains in the Netherlands.

Figure 4
View larger version:
Figure 4

Five largest clusters (A—E) in the Netherlands, 1993—1997: cases by time of diagnosis and risk group.

Risk factors for active transmission of tuberculosis

Demographic factors associated with clustering of cases were male sex, young age, nationality, urban residence, and residence time in the Netherlands (table 1). Females were less likely to be in a cluster than males. Increasing age was strongly associated with decreasing clustering. Among the non-Dutch, with the exception of Surinamese patients, a smaller proportion of cases was clustered than among the Dutch. The proportion of clustered cases increased with time of residence in the Netherlands (χ2trend, P < .0001). In the four largest cities, a slightly larger proportion of tuberculosis cases was clustered than elsewhere. Tuberculosis patients who had been traveling to highly endemic areas were less frequently clustered. Clustering of cases was more common among drug users and tended to be increased among homeless people (OR, 1.8), though this was not statistically significant in multivariate analysis. HIV seropositivity was not found to be a risk factor for clustering (OR, 1.2; 95% confidence interval [CI], 0.9–1.6). Extrapulmonary cases were less likely to be in a cluster than pulmonary cases. Patients who had been previously treated for tuberculosis were somewhat less likely to be in a cluster. As described above, case detection through contact tracing was strongly associated with clustering. Patients with isoniazid-resistant strains were less likely to be in a cluster, also after adjusting for confounders. However, for the Dutch this was not significant (P < .05).

Table 1
View larger version:
Table 1

Risk factors for clustering of tuberculosis (TB) cases in the Netherlands, 1993–1997.

Transmission between immigrants and Dutch population.

To identify risk factors associated with transmission between Dutch and non-Dutch cases, clustered patients were subdivided into clusters with Dutch patients only, into mixed clusters composed of Dutch-non-Dutch patients, and into those with only non-Dutch patients. Young and middle age, urban residence, drug abuse, and isoniazid resistance were risk factors associated with Dutch patients' being in mixed clusters (table 2). It may be worth noting that the HIV-infected Dutch patients were mainly found among drug users, and, therefore, adjustment for drug use strongly decreased the association with HIV infection (adjusted OR, 1.5; 95% CI, 0.6–3.8). Risk factors among non-Dutch patients for being in a cluster containing Dutch patients were middle age, Surinamese nationality, duration of stay in the Netherlands, urban residence, drug use, and pulmonary localization of tuberculosis (table 3).

Table 2
View larger version:
Table 2

Risk factors for Dutch patients in a mixed Dutch-non-Dutch cluster.

Table 3
View larger version:
Table 3

Risk factors for non-Dutch patients in a mixed Dutch-non-Dutch cluster.

Transmission of drug-resistant and drug-susceptible tuberculosis.

Isoniazid-resistant strains were less likely to be in a cluster than isoniazid-sensitive strains (OR, 0.7; 95% CI, 0.5–0.9; table 1). In table 4, data are presented separately for the Dutch and non-Dutch. Among non-Dutch patients, 36% of isoniazid-resistant strains were clustered, compared with 45% of isoniazid-sensitive strains (crude and adjusted OR, 0.7). Similarly, among the Dutch, the crude and adjusted ORs were 0.7, and 0.4, respectively.

Table 4
View larger version:
Table 4

Drug resistance and clustering of restriction fragment length polymorphism patterns in the Netherlands, 1993–1997.

Previous SectionNext Section

Discussion

The findings of this study show that the percentage of clustering is strongly dependent on the age of the patients in a study population, size of the study population, and length of the study period. Therefore, the degree of clustering in the various studies published cannot be interpreted directly in terms of the degree of ongoing transmission, even when the values of clustering obtained appear to be similar. The proportion of cases clustered in our study was 46%, and the proportion attributable to active transmission was estimated to be 35%. These values are comparable with those in studies conducted in San Francisco [4], New York [5], Zurich [35], Bern [36], Denmark [14], and South Africa [15]. However, a major difference with these studies is the extended study period of 5 years and large sample size, 4266 patients, in the present study. The sample size and percentage of clustered cases varied: approximately 100 cases in New York (37.5% clustered) [5], 246 in South Africa (45%) [15], 500 in San Francisco (40%) [4], and 1500 in Denmark (50%) [14]. As a first attempt to enable comparison between molecular epidemiological data from different regions, one may plot the percentage of clustered cases against the sample size. The gradient of this graph may be a crude marker for ongoing transmission rather than the fraction of clustered cases for the whole target population. In the present study, the percentage of clustered cases amounted to 11%, 19%, 23%, and 35% for the sample sizes of 100, 200, 400, or 1000 consecutive cases. These values suggest significant less ongoing transmission in the Netherlands compared with any other geographical region studied thus far. This low value is consistent with the nature of the population studied in the Netherlands. All other studies except the Danish one were done in large cities, and our data indicate increased clustering in large cities.

In previous studies, the proportion of cases clustered was found to be highest at young ages [4, 5, 13]. The present study confirms these observations but suggests different underlying mechanisms among the Dutch and non-Dutch. Among the Dutch, an increased incidence of nonclustered tuberculosis at older ages (145 years of age) is of major importance, whereas among the non-Dutch a decreasing incidence of clustered tuberculosis is seen with increasing age (⩽55 years of age). For the Dutch, this might be due to a strongly increasing infection prevalence with increasing age. This is due to a cohort effect: the older population was exposed to a high risk of infection when they were young [37]. Among the non-Dutch, this cohort effect is expected to be much weaker. In addition, the association between age and clustering among immigrants is confounded by recent immigration, since this is associated both with less clustering and with young age.

Outbreak investigations have shown that M. tuberculosis isolates from epidemiologically related tuberculosis patients are usually identical. However, the evidence of epidemiological linkage among patients clustered by matching RFLP patterns in population-based studies is not well established. Also in this study, the evidence that linkage by RFLP matching represents an epidemiological linkage was only circumstantial: (1) virtually all tuberculosis cases disclosed by active case finding from contact tracing were clustered, and in 91% of these cases evidence for epidemiological linkage was found; (2) regional public health institutes reported an epidemiological connection among 45% of the clustered tuberculosis cases; and (3) clustering was found to be associated with known risk factors for transmission, such as young age, drug abuse, and pulmonary tuberculosis.

It should be noted that our analysis may have overestimated active transmission due to the circulation of M. tuberculosis strains with stable RFLP patterns in the population or underestimated active transmission due to rapidly changing IS6110 patterns. In a recent study, Yeh et al. found that about 30% of serial isolates of patients whose cultures spanned at least 90 days had changed IS6110 RFLP patterns [38]. Preliminary investigations in our laboratory confirmed that IS6110 RFLP types have a relatively short half-life time, which may be in the range of a few years (A. de Boer, unpublished data). Therefore, it seems likely that the degree of ongoing transmission estimated in our study with a time window of 4 years should be corrected for some underestimation because of IS6110 instability during this period of time, which is longer than in most other studies.

More than half of the clustered patients in this study were in clusters composed of ⩽5 individuals, and only a minority (19%) were in large clusters composed of 15 or more cases. The cluster size distribution in our study was very different from Denmark, where 40% of clustered patients were found in only 2 clusters [14]. The 5 largest clusters in the Netherlands was composed of only 9.5% of the clustered patients. Our study confirms the observation of others, that patients in large clusters share sociodemographic characteristics, such as homelessness, drug abuse, and ethnicity [3]. An exception was the largest cluster (cluster A), composed of 47 cases, because most cases were among low-risk individuals. The latter microepidemic was initiated by a chronic excretor whose disease was not recognized for many months. This example reemphasizes that failure to timely detect and treat a single chronic infectious patient may have a significant impact on the number of newly acquired infections in a population [4].

The occurrence of large clusters may indicate weaknesses in the tuberculosis control activities, as they were still growing at the end of 1997. Transmission among the Somali immigrants may have occurred partly before their arrival in the Netherlands. However, 80% of cases in this cluster were identified ⩾2 years after arrival in the Netherlands, suggesting that transmission in the Netherlands may have occurred as well. Contact tracing is reportedly difficult among Somali immigrants because of cultural and language barriers. The major intervention in this group may be to promote increased compliance with tuberculosis screening and/or increased health education to ensure early case detection when symptoms occur. Contact tracing among drug users and the homeless is notably difficult as well, and additional measures to prevent tuberculosis transmission in this group and the need to promote early case detection are again emphasized [17].

Risk factors for clustering as disclosed in this study, such as male gender, young age, homelessness, drug abuse, and a long stay in the country of residence, have previously been described in other studies [4, 14]. Similar to findings in the United States, we found somewhat more transmission in urban environments, in particular among drug abusers and the homeless, and also between immigrants and the native population [17, 20, 21]. Pulmonary cases were more likely to be clustered than extrapulmonary tuberculosis cases, which is to be expected, as pulmonary cases are the main sources of infection [37] and are thus able to generate clusters. Contrary to studies in the United States [4, 5], HIV was not found to be associated with clustering after adjusting for drug use. Perhaps the association in the United States occurred in part because of nosocomial transmission in HIV/AIDS clinics [18, 39].

Studies in the United States showed that drug-resistant M. tuberculosis cases were more likely to be in a cluster than drug-sensitive cases [4, 5, 39–41]. In our study population, isoniazid- resistant strains were less likely to be clustered than isoniazid-sensitive strains or streptomycin-resistant strains, as had been suggested in an earlier, preliminary report [26]. Moreover, in the present study, potential confounders were taken into account, and the association remained. It has been known for decades that isoniazid resistance may lead to reduced virulence of M. tuberculosis strains in animal models [42, 43]. This study indicates on the basis of DNA cluster analysis that isoniazid-resistant strains may be less likely to generate secondary cases in a human population. Preliminary data suggest that the prevalence of particular mutations in the katG gene of isoniazid differ between clustered and nonclustered resistant strains. This would suggest differences in transmissibilty of the various genotypes of isoniazid resistance. If confirmed, one would expect that the prevalence of various resistance genotypes in a population may be exploited as an indicator of the prevalence of acquired versus primary resistance in a population.

  • Received November 12, 1998.
  • Revision received April 26, 1999.
Previous Section
 

References

  1. 1.↵
    1. Daley CL,
    2. Small PM,
    3. Schechter GF,
    4. et al
    . An outbreak of tuberculosis with accelerated progression among persons infected with the human immunodeficiency virus. N Engl J Med 1992;326:231-5.
    MedlineWeb of Science
  2. 2.
    1. Kline SE,
    2. Hedemark LL,
    3. Davies SF
    . Outbreak of tuberculosis among regular patrons of a neighborhood bar. N Engl J Med 1995;333:222-7.
    CrossRefMedlineWeb of Science
  3. 3.↵
    1. Kiers A,
    2. Drost AP,
    3. Van Soolingen D,
    4. Veen J
    . Use of DNA fingerprinting in international source case finding during a large outbreak of tuberculosis in The Netherlands. Int J Tuberc Lung Dis 1997;1:239-45.
    MedlineWeb of Science
  4. 4.↵
    1. Small PM,
    2. Hopewell PC,
    3. Singh SP,
    4. et al
    . The epidemiology of tuberculosis in San Francisco. N Engl J Med 1994;330:1703-9.
    CrossRefMedlineWeb of Science
  5. 5.↵
    1. Alland D,
    2. Kalkut GE,
    3. Moss AR,
    4. et al
    . Transmission of tuberculosis in New York City. N Engl J Med 1994;330:1710-6.
    CrossRefMedlineWeb of Science
  6. 6.↵
    1. Van Deutekom H,
    2. Gerritsen JJJ,
    3. Van Soolingen D,
    4. Van Ameijden EJC,
    5. Van Embden JDA,
    6. Coutinho RA
    . Molecular epidemiological approach to studying the transmission of tuberculosis in Amsterdam. Clin Infect Dis 1997;25:1071-7.
    Abstract/FREE Full Text
  7. 7.↵
    1. Small PM,
    2. McCleeny NB,
    3. Singh SP,
    4. Schoolnik GK,
    5. Thompkins LS,
    6. Mickelsen PA
    . Molecular strain typing of Mycobacterium tuberculosis to confirm cross-contamination in the mycobacteriology laboratory and modification of procedures to minimize occurrence of false-positive cultures. J Clin Microbiol 1993;31:1677-82.
    Abstract/FREE Full Text
  8. 8.
    1. Braden CR,
    2. Templeton GL,
    3. Stead WW,
    4. Bates JH,
    5. Cave MD,
    6. Valway SE
    . Retrospective detection of laboratory cross-contamination of Mycobacterium tuberculosis cultures with use of DNA fingerprint analysis. Clin Infect Dis 1997;24:35-40.
    Abstract/FREE Full Text
  9. 9.↵
    1. Van Duin JM,
    2. Pijnenburg JEM,
    3. Van Rijswoud CM,
    4. De Haas PEW,
    5. Hendriks WDH,
    6. Van Soolingen D
    . Investigation of cross contamination in a Mycobacterium tuberculosis laboratory using IS6110 DNA fingerprinting. Int J Tuberc Lung Dis 1998;2:425-9.
    MedlineWeb of Science
  10. 10.↵
    1. Borgdorff MW,
    2. Nagelkerke N,
    3. Van Soolingen D,
    4. De Haas PEW,
    5. Veen J,
    6. Van Embden JDA
    . Analysis of tuberculosis transmission between nationalities in The Netherlands in the period 1993–1995 using DNA fingerprinting. Am J Epidemiol 1998;147:187-95.
    Abstract/FREE Full Text
  11. 11.↵
    1. Canetti G,
    2. Sutherland I,
    3. Svandova E
    . Endogenous reactivation and exogenous reinfection. Their relative importance with regard to the development of non-primary tuberculosis. Bull Int Union Tuberc 1972;47:123-4.
  12. 12.↵
    1. Vynnycky E,
    2. Fine PEM
    . The natural history of tuberculosis: the implications of age-dependent risks of disease and the role of reinfection. Epidemiol Infect 1997;119:183-201.
    CrossRefMedline
  13. 13.↵
    1. Yang ZH,
    2. De Haas PEW,
    3. Wachman CH,
    4. Van Soolingen D,
    5. Van Embden JDA,
    6. Anderson AB
    . Molecular epidemiology of tuberculosis in Denmark in 1992. J Clin Microbiol 1995;33:2077-81.
    Abstract/FREE Full Text
  14. 14.↵
    1. Bauer J,
    2. Yang Z,
    3. Poulsen S,
    4. Andersen AB
    . Results from 5 years of nationwide DNA fingerprinting of Mycobacterium tuberculosis complex isolates in a country with a low incidence of M. tuberculosis infection. J Clin Microbiol 1998;36:305-8.
    Abstract/FREE Full Text
  15. 15.↵
    1. Wilkinson D,
    2. Pillay M,
    3. Crump J,
    4. Davies GR,
    5. Sturm AW
    . Molecular epidemiology and transmission dynamics of Mycobacterium tuberculosis in rural Africa. Trop Med Int Health 1997;2:747-53.
    CrossRefMedlineWeb of Science
  16. 16.↵
    1. Glynn JR,
    2. Vynnycky E,
    3. Fine PEM
    . The influence of incomplete case ascertainment on estimates of recent transmission of M. tuberculosis using DNA fingerprinting technique. Int J Tuberc Lung Dis 1997;1(Suppl):S94.
  17. 17.↵
    1. Barnes PF,
    2. Yang Z,
    3. Preston-Martin S,
    4. et al
    . Patterns of tuberculosis transmission in Central Los Angeles. JAMA 1997;278:1159-63.
    Abstract/FREE Full Text
  18. 18.↵
    1. Edlin BR,
    2. Tokars JI,
    3. Grieco MH
    . An outbreak of multidrug-resistant tuberculosis among hospitalized patients with the acquired immunodeficiency syndrome. N Engl J Med 1992;23:1514-21.
  19. 19.
    1. Burman WJ,
    2. Reves RR,
    3. Hawkes AP,
    4. et al
    . DNA fingerprinting with two probes decreases clustering of Mycobacterium tuberculosis. Am J Respir Crit Care Med 1997;155:1140-6.
    Abstract/FREE Full Text
  20. 20.↵
    1. Tornieporth NG,
    2. Ptachewich Y,
    3. Poltoratskaia N,
    4. et al
    . Tuberculosis among foreign-born persons in New York City, 1992–1994: implications for tuberculosis control. Int J Tuberc Lung Dis 1997;1:528-35.
    MedlineWeb of Science
  21. 21.↵
    1. Jasmer RM,
    2. Ponce de Leon A,
    3. Hopewell PC,
    4. et al
    . Tuberculosis in Mexican-born persons in San Francisco: reactivation, acquired infection and transmission. Int J Tuberc Lung Dis 1997;1:536-41.
    MedlineWeb of Science
  22. 22.↵
    1. Braden CR,
    2. Templeton GL,
    3. Cave MD,
    4. et al
    . Interpretation of restriction fragment length polymorphism analysis of Mycobacterium tuberculosis isolates from a state with a large rural population. J Infect Dis 1997;175:1446-52.
    Abstract/FREE Full Text
  23. 23.↵
    1. Raviglione MC,
    2. Sudre P,
    3. Rieder HL,
    4. Spinaci S,
    5. Kochi A
    . Secular trends of tuberculosis in Western Europe. Bull World Health Organ 1993;71:297-306.
    MedlineWeb of Science
  24. 24.↵
    1. Veen J,
    2. Kalisvaart NA
    . Index Tuberculosis 1996 (In Dutch). The Hague: Royal Netherlands Tuberculosis Association; 1997.
  25. 25.↵
    WHO book on drug resistance surveillance. Geneva: World Health Organization; 1997.
  26. 26.↵
    1. Van Soolingen D,
    2. Lambregts-van Weezenbeek CSB,
    3. de Haas P,
    4. Veen J,
    5. van Embden JDA
    . Transmissie van gevoelige en resistente Mycobacterium tuberculosis stammen in Nederland, 1993–1995, onderzocht met DNA fingerprinting. Ned Tijdschr Geneeskd 1996;140:2286-9.
    Medline
  27. 27.↵
    1. Lambregts-van Weezenbeek CSB,
    2. Jansen HM,
    3. Veen J,
    4. Nagelkerke NJD,
    5. Sebek MMGG,
    6. van Soolingen D
    . Origin and management of primary and acquired drug-resistant tuberculosis in The Netherlands: the truth behind the rates. Int J Tuberc Lung Dis 1998;2:296-302.
    MedlineWeb of Science
  28. 28.↵
    1. Sebek MMGG
    . Transmission of M. tuberculosis in The Netherlands, results December 1994–December 1996. The surveillance project of DNA fingerprinting. Int J Tuberc Lung Dis 1997;1(Suppl):S57.
  29. 29.↵
    1. Gangadharam PRJ
    . Drug resistance in Mycobacteria. Boca Raton, FL: CRC Press; 1984.
  30. 30.↵
    1. Van Embden JDA,
    2. Cave MD,
    3. Crawford JT,
    4. et al
    . Strain identification of Mycobacterium tuberculosis by DNA fingerprinting: recommendations for a standardized methodology. J Clin Microbiol 1993;31:406-9.
    Abstract/FREE Full Text
  31. 31.↵
    1. Van Soolingen D,
    2. De Haas PEW,
    3. Hermans PWM,
    4. Groenen P,
    5. Van Embden JDA
    . Comparison of various repetitive DNA elements as genetic markers for strain differentiation and epidemiology of M tuberculosis. J Clin Microbiol 1993;31:1987-95.
    Abstract/FREE Full Text
  32. 32.↵
    1. Ross C,
    2. Raios K,
    3. Jackson K,
    4. Dwyer B
    . Molecular cloning of a highly repeated element from Mycobacterium tuberculosis and its use as an epidemiological tool. J Clin Microbiol 1992;30:942-6.
    Abstract/FREE Full Text
  33. 33.↵
    1. Parish T.,
    2. Stoker N.G.
    1. Heersma HF,
    2. Kremer K,
    3. van Embden JDA
    . Computeranalysis of RFLP patterns of Mycobacterium tuberculosis, chap. 29. In: Parish T., Stoker N.G., editors. Methods in molecular biology. Vol. 101. Totowa, NJ: Humana Press; 1998.
  34. 34.↵
    1. Van Soolingen D,
    2. Qian L,
    3. De Haas PEW,
    4. et al
    . Predominance of a single genotype of Mycobacterium tuberculosis in countries of East Asia. J Clin Microbiol 1995;33:3234-8.
    Abstract/FREE Full Text
  35. 35.↵
    1. Pfyffer GE,
    2. Strässle A,
    3. Rose N,
    4. Wirth R,
    5. Brändli O,
    6. Shang H
    . Tuberculosis in the Metropolitan Area of Zurich; A 3-year survey based on DNA Fingerprinting. Eur Respir J 1998;11:804-8.
    Abstract
  36. 36.↵
    1. Genewein A,
    2. Telenti A,
    3. Bernasconi C,
    4. et al
    . Molecular approach to identifying route of transmission of tuberculosis in the community. Lancet 1993;342:841-4.
    CrossRefMedlineWeb of Science
  37. 37.↵
    1. Styblo K
    . Epidemiology of tuberculosis. Selected papers no 24. The Hague: Royal Netherlands Tuberculosis Association; 1991.
  38. 38.↵
    1. Yeh RW,
    2. Ponce-de-Leon A,
    3. Agasino CB,
    4. et al
    . Stability of Mycobacterium tuberculosis DNA genotypes. J Infect Dis 1998;177:1107-11.
    Abstract/FREE Full Text
  39. 39.↵
    1. Frieden TH,
    2. PI Fujiwara,
    3. RM Washko,
    4. MA Hanburg
    . Tuberculosis in New York city—turning the tide. N Engl J Med 1995;333:229-33.
    CrossRefMedlineWeb of Science
  40. 40.
    1. Snider DE Jr.,
    2. Kelly GD,
    3. Cauthen GM,
    4. Thompson NJ,
    5. Kilburn JO
    . Infection and disease among contacts of tuberculosis cases with drug-resistant and drug-susceptible bacilli. Am Rev Respir Dis 1985;132:125-32.
    MedlineWeb of Science
  41. 41.↵
    1. Beck-Sagué C,
    2. Dooley SW,
    3. Hutton MD,
    4. et al
    . Hospital outbreak of multi-drug-resistant Mycobacterium tuberculosis infections: factors of transmission to staff and HIV-infected patients. JAMA 1992;268:1280-6.
    Abstract/FREE Full Text
  42. 42.↵
    1. Middlebrook G
    . Isoniazid resistance and catalase activity of tubercle bacilli. Am Rev Tuberc 1954;69:471-2.
    MedlineWeb of Science
  43. 43.↵
    1. Wolinsky E,
    2. Smith MM,
    3. Steenken W
    . Isoniazid susceptibility, catalase activity, and guinea pig virulence of recently isolated cultures of tubercle bacilli. Am Rev Tuberc 1956;73:768-72.
    MedlineWeb of Science
| Table of Contents

This Article

  1. J Infect Dis. (1999) 180 (3): 726-736. doi: 10.1086/314930
  1. AbstractFree
  2. » Full Text (HTML)Free
  3. Full Text (PDF)Free

Classifications

Share

  1. Email this article

Navigate This Article

Current Issue

  1. March 1, 2012 205 (5)
  1. Journal of Infectious Diseases
  1. Alert me to new issues

Most