The traditional treatment of tuberculosis (TB) infection (9 months of daily isoniazid [9H]) is safe but completion rates of <50% are reported. Shorter regimens (3 months of once-weekly isoniazid and rifapentine [3HP] or 4 months of daily rifampin [4R]) are associated with improved adherence in adults.
This was a retrospective cohort study (2014–2017) of children (0–18 years old) seen at a children’s TB clinic in a low-incidence nation. We compared the frequency of completion and adverse events (AEs) in children receiving 3HP, 4R, and 9H; the latter 2 regimens could be administered by families (termed self-administered therapy [SAT]) or as directly observed preventive therapy (DOPT); 3HP was always administered under DOPT.
TB infection treatment was started in 667 children: 283 (42.4%) 3HP, 252 (37.8%) 9H, and 132 (19.8%) 4R. Only 52% of children receiving 9H via SAT completed therapy. Children receiving 3HP were more likely to complete therapy than the 9H (SAT) group (odds ratio [OR] 27.4, 95% confidence interval [CI]: 11.8–63.7). Multivariate analyses found receipt of medication under DOPT (OR: 5.72, 95% CI: 3.47–9.43), increasing age (OR: 1.09, 95% CI: 1.02–1.17), and the absence of any AE (OR: 1.70, 95% CI: 0.26–0.60) to be associated with completing therapy. AEs were more common in the 9H group (OR: 2.51, 95% CI: 1.48–4.32). Two (0.9%) children receiving 9H developed hepatotoxicity; no child receiving 3HP or 4R developed hepatotoxicity.
Shorter regimens are associated with increased completion rates and fewer AEs than 9H.
Treatment of tuberculosis infection is safe and well tolerated. Completion rates of 9 months of daily isoniazid (9H) are <50%. Using shorter regimens (3 months of once-weekly isoniazid and rifapentine [3HP] and 4 months of daily rifampin) can enhance completion.
Children receiving 3HP were more likely to complete therapy than those receiving daily 9H (odds ratio: 27.4, 95% confidence interval: 11.8–63.7); only 52% of those receiving daily 9H completed therapy. Hepatotoxicity was uncommon (0% in 3HP vs 0.9% in 9H).
Therapy for tuberculosis (TB) infection is safe, efficacious, and recommended for children who receive test results that are positive for an infection (either the tuberculin skin test [TST] or an interferon γ release assay [IGRA]) unless specific contraindications exist.1 In contrast, effectiveness (how an intervention performs in real-world conditions) has been poor. The long duration of therapy with the US-recommended 9 months of daily isoniazid (9H), concerns that families have regarding false-positive TSTs in children who have received the BCG vaccine, and a culture of TB prevention that has made few inroads into high-prevalence settings have adversely impacted the effectiveness of TB infection therapy. These factors may be particularly important for immigrant children, a group in which the rate of TB infection is far higher than for US-born children.2 Authors of a recent study found that only 12% of over 8200 children diagnosed with TB infection before immigrating to the United States successfully completed therapy after arrival in the United States.3
Completion rates of <50% have been reported for both children and adults when TB medications are administered by families.4,5 There are few patient-specific factors associated with completion of therapy. Our previous studies found that region of origin, language, and the method to test for TB infection (TST versus IGRA) were not associated with completion of therapy.6 However, provision of therapy by local health departments7 and using shorter-course therapy with 4 months of daily rifampin (4R)7 or 3 months of once-weekly isoniazid and rifapentine (3HP)8 were associated with completion rates of 93% to 99%.
The last decade has seen publication of more robust data on IGRA use in children9 and on shorter course regimens such as 3HP.10,11 The ability to more accurately identify children who would benefit from therapy to prevent progression to disease, coupled with regimens more palatable to children and their families, suggests strategies to improve effectiveness of TB infection therapy. The goal of this study was to describe the trends of use, completion rates, and safety profiles of the 3 most commonly used regimens for TB infection in a heterogeneous cohort of children in a busy US urban TB clinic.
This retrospective cohort study included 0- to 18-year-old children who were offered treatment for TB infection from January 2014 to March 2017 at a children’s TB clinic associated with a quaternary care children’s hospital. This clinic is the main referral venue for immigration and refugee programs, public health departments, and a network of community pediatricians in a metropolitan area of almost 7 million persons. Using American Thoracic Society and Centers for Disease Control and Prevention guidelines, TB infection was defined as having positive results from either a TST or IGRA.12,13 Children receiving regimens for suspected multidrug-resistant TB infection, families who refused to start children on therapy, or those still on therapy were excluded from this study. Eighty children previously published8 as receiving 3HP were included.
All TSTs were performed before children were being seen in the TB clinic. A positive TST result alone (without a confirmatory sequential IGRA) defined TB infection in: (1) children identified through contact tracing of a person with TB disease, (2) immunocompromised children, (3) children about to receive immunosuppressive therapy, (4) children <2 years of age, (5) children with TST induration of ≥20 mm, and (6) children with a positive TST after having previously negative TST results. For other children, positive TST results were followed by an IGRA obtained in our clinic and treatment was offered only if the IGRA result was positive for TB. The rationale behind why our community colleagues used TST versus IGRA was not available.
All children were treated by the authors using 1 of 3 regimens (Table 1). 3HP was used in children ≥2 years old and always administered as directly observed preventive therapy (DOPT).10,11 Children <2 years old were not prescribed this regimen given the absence of data on rifapentine dosing in this group.14 The remaining regimens, 4R and 9H, were used in children of all ages and could be administered as DOPT, enhanced self-administered therapy ([ESAT] in which medications were delivered monthly to families by health departments, who called periodically for reminders; daily medication was administered by family members), or self-administered therapy (SAT). In general, all 3 options were presented to patients ≥2 years old, but a preference was expressed by providers for 3HP. Isoniazid (INH) was only administered twice-weekly under DOPT. In DOPT, medications were delivered by a health department representative who asked the family about symptoms before administering the subsequent dose. Completion was calculated for the first regimen a child received. An exception was if therapy had to be changed because of documented drug resistance from an isolate from a person close to the child; then, it was that definitive regimen that was used to evaluate completion.
The primary outcome was completion of therapy. Children were considered to have completed therapy (Table 1) if medication was given under DOPT, even if they did not return for clinic visits as long as medication receipt was documented by the health department. For children returning for follow-up, caregivers were questioned about medication administration. Families not returning for follow-up were called to inquire about adherence and medication tolerance. Subjects receiving medication by SAT who did not return for follow-up or could not be contacted were considered to have not completed therapy.
Our secondary outcome was safety and tolerability. Children were seen in clinic at the initiation of therapy, 4 to 6 weeks after starting therapy, and every 1 to 3 months thereafter. We did not routinely obtain liver function tests (LFTs) in otherwise healthy children. Children receiving other potentially hepatotoxic medications and children with obesity (in whom nonalcoholic fatty liver disease could lead to baseline transaminitis) did have LFTs obtained at baseline. Children with abdominal pain, appetite or weight loss, vomiting, or icterus had medication(s) withheld pending LFTs being obtained. Medical records were abstracted for demographic variables, regimen doses, adherence, and adverse events (AEs). Aspartate and alanine aminotransferase levels were abstracted. AEs were graded using the National Cancer Institute’s severity grading system.15 Institutional review board approval was obtained before study onset. Stata 11 (StataCorp, College Station, TX) was used for analyses.
During the study period, 667 children were treated for TB infection. Two families (0.3% of all families seen) refused to initiate therapy; the remainder all agreed to start therapy at the time of the first clinic visit. Children were born in 54 countries and families spoke 25 different languages; in-person or phone translation was used. The most common reasons for testing for TB infection were as follows: 330 (49.5%) birth in or 59 (8.8%) travel to high-prevalence countries, 211 (33.1%) public health contact tracing, and 19 (2.8%) before immunosuppression or as a prelude to organ transplantation (Table 2). The only variable associated with selection of regimens was age; there were no significant differences in the use of regimens based on other demographic variables, comorbid conditions, or reason for testing for TB infection.
Shift in Diagnosis and Treatment
Most children were tested for TB infection by using TSTs before arrival in our clinic. During the study period, use of IGRAs, either in isolation or obtained after a child had a positive TST results, increased (P value for trend in odds <.0001) (Fig 1A). IGRA use was higher among children ≥5 years old: 246 out of 438 (56.2%) children had IGRAs performed compared with 34 out of 229 (14.8%) children <5 years of age (odds ratio [OR]: 7.3, 95% confidence interval [CI]: 4.9–11.1). The most common treatment regimen was 3HP (283, 42.4%), followed by 9H (252, 37.8%), and 4R (132, 19.8%). Use of 3HP increased and 9H decreased significantly over the study period (P < .0001) (Fig 1B). Children who received 9H were significantly younger than those who received 3HP or 4R (mean 4.5 versus 11.2 years, P < .0001). Even when children <2 years were excluded, the mean age of 9H recipients remained younger than children receiving 3HP or 4R (mean 5.8 versus 11.5 years, P < .0001).
The highest completion rates were seen with 3HP (96.8%) and the lowest with 9H under SAT (52.6%) (OR: 27.4, 95% CI: 11.8–63.7) (Table 3). 9H demonstrated completion rates of 89% (in 158 out of 178 children) when administered under DOPT, but completion was only 53% (in 30 out of 57 children) when administered under SAT (OR: 7.11, 95% CI: 3.54–14.28). There was no difference in completion rates between 4R under DOPT and 3HP (OR: 1.12, 95% CI: 0.14–9.09). On univariate analyses, completion was more common in older (≥2 year old) children compared with <2-year-olds (OR: 2.37, 95% CI: 1.30–4.32) and for children identified via contact investigations versus children with other risk factors for TB infection. When subgroup analyses were performed on children ≥2 years old, completion rates were significantly higher in the 3HP group (274 out of 283 children, 96.8%) than in the 9H group (148 out of 184 children, 80.4%), (OR: 7.4, 95% CI 3.5–15.8). There was no difference in completion rates under DOPT for children identified through contact investigations versus those tested for TB infection for other reasons (OR: 0.80, 95% CI: 0.38–1.68).
Failure to complete therapy was associated with the development of any AE (OR: 0.1, 95% CI: 0.06–0.2) and with therapy administered outside of DOPT (Table 3). A regression model for completion included age, race and/or ethnicity, child and/or family region of residence, BCG status, test used for TB infection, mode of administration, regimen used, and the presence of any AE. Receipt of medication by DOPT (OR: 5.72, 95% CI: 3.47–9.43), increasing age (OR: 1.09, 95% CI: 1.02–1.17), and the absence of any AE (OR: 1.70, 95% CI: 1.43–2.02) were associated with completing therapy. Administration of medication by families (OR: 0.40, 95% CI: 0.26–0.60) was associated with failure to complete therapy. As the study period proceeded, more children were likely to complete therapy for TB infection (P value for trend of odds = .0029), due at least in part to increased use of 3HP and DOPT.
Sixty-two (9.3%) children complained of any AE (Table 4). Complaints of any AE were more common in the 9H group compared with 3HP and 4H groups (OR: 2.51, 95% CI: 1.48–4.32). There was no difference in hepatotoxicity among the regimens (P = .192). The most AE common was abdominal pain (41 out of 667 children, 6.1%); only 2 out of 667 children treated (0.3%) had elevated hepatic transaminases and the remainder had normal transaminase levels. Only 1 child had a grade 4 AE: a 2-year-old girl with nephrotic syndrome who developed aspartate 1115 U/L and alanine aminotransferase 1490 U/L in her eighth month of INH. Therapy was immediately stopped and she recovered. One child, a 15-year-old with lupus and idiopathic thrombocytopenic purpura, died of a hemorrhagic stroke while on 4R; she had previous strokes before starting 4R, and her stroke was thought to be because of her underlying diseases. A 19-month-old child developed hematochezia as a result of a Meckel’s diverticulum while on 9H; he was not coagulopathic.
Nineteen (2.8%, 30.6% of all children with any AE) patients required a change in regimen because of AEs. Seven families refused a second regimen and 12 families changed therapy; 7 from 9H to 4R, 4 from 3HP to 4R, and 1 from 4R to 9H. The most common reasons for changing therapy were abdominal pain (8), rash (4), and spitting out medication with the DOT worker (4). In 10 out of 19 (52.6%) instances, children successfully completed the second regimen, 2 children had INH stopped after at least 6 months had been given, and the remaining families opted to not restart another regimen.
A 16-year-old who received 3HP developed culture-confirmed pulmonary TB 7 months after completing therapy. There was concern that she hid her medications in her cheek and spit them out after the health department worker departed; her isolate was pan susceptible. We are the referral center for all children in an 8-county area with TB disease and we were not informed by any of the health departments that any of our other children developed TB disease after therapy for TB infection.
Low rates of completion of 9H have been extensively reported.3,–5 Unfortunately, this has not greatly changed practice patterns, with over 85% of pediatric infectious disease physicians continuing to recommend 9H as first-line therapy.16 Pediatricians in general practice, who treat most children with TB infection, may feel less comfortable moving away from 9H because this is the regimen with which they have the most experience. We found that despite extensive education on the importance of treatment and tolerability of medications, only 50% completed therapy with 9H administered by families.6 This led us to be early adopters of shorter course regimens and advocates of administration under DOPT. We wanted to share the results of our experience with 9H and shorter-course regimens generated from our high-volume referral TB clinic to demonstrate postmarketing data on completion rates and safety. There were several pertinent findings. First, completion rates were significantly higher for shorter-course regimens than 9H, when medications were administered by families. Second, short-course regimens were well tolerated. Third, effectiveness is high in the short-term.
Successful completion of therapy was not associated with demographic factors in our study. This means that choices that clinicians make have the greatest impact on completion of therapy and reduction of TB disease in the future. Administration of shorter-course regimens, administration of therapy by local health departments, or both, have the greatest impact on whether a child completes therapy. The role of DOPT is not solely in terms of medication administration, although this is a critical role that helps overcome the barrier of a family’s ability to purchase medications and obtain refills. The other roles of DOPT are to assess for AEs on a regular basis and to encourage the child and family to take medications. Many families that were nonadherent with clinic visits nonetheless completed therapy under DOPT because health departments worked with families’ schedules and these visits did not interfere with parental work schedules. Thus, although we were unable to see many children on the recommended schedule, we were still apprised immediately if children had any AEs.
Because DOPT is not available throughout the United States, providers should consider moving from 9H to 4R if therapy is to be administered by families. Family administration of rifampin (RIF) was associated with far higher completion rates than 9H. The favorable safety profile, the ability to retain once-daily dosing, and the relative ease of acquisition of RIF in the community facilitate uptake of this regimen. Another alternative is to have more robust capacities for remotely monitoring adherence and assessing toxicity, as has been successfully accomplished with video DOPT programs. In video DOPT, medications are left for families who administer the medication to children in front of a phone-based video application for smart phones that allows for secure transmission of the video file to the health department. Video DOPT is associated with similar completion rates as to in-person DOPT and with lower costs.17
Short-course therapy was safe for children. Based on previous studies of combination therapy for TB infection with RIF and pyrazinamide,18 there was concern that use of >1 anti-TB drug could result in hepatotoxicity. This was not borne out in 3HP studies, in which hepatotoxicity rates were over 4 times lower for 3HP than for 9H19; this is thought to be because of a reduction in the total exposure time to hepatically metabolized medications. We observed no cases of hepatotoxicity in our patients who received 3HP and a low frequency of hepatotoxicity in children receiving 9H. Despite actively asking families about systemic drug reactions, we did not note many of the systemic drug reactions with 3HP, including influenza-like illnesses and rashes, that have been reported up to 63% and 17% of adults, respectively.20 Myalgias were self-limited, predominantly occurred in the third to fourth week of therapy, as has been described in adults, and were responsive to nonsteroidal anti-inflammatory drugs. When adolescents were given the option of continuing 3HP or stopping therapy and beginning therapy anew with a longer regimen, all opted to remain on 3HP.
Short-course therapy was effective, at least in the short-term. Our 3HP data mirrors that of the large prospective trial that enrolled children 2 to 17 years of age,10 but given the short follow-up period, we cannot comment on the longer-term effectiveness of 3HP in our patients. However, the vast majority of children who develop TB disease do so within 2 years, which increases the likelihood that the treatment was truly effective. Given the increasing use of IGRAs in our clinic and our recommendation to community providers to order IGRAs more frequently as a testing tool, we feel that these children were truly infected, and that their positive test results were not false-positive reactions caused by non-TB mycobacteria or previous BCG vaccination. The improved specificity of IGRAs, particularly in the child >2 years of age, enables providers to target resources to children who are most likely to receive benefit from therapy.
There were limitations to this study. There were a small number of missing data elements in this retrospective study. Patients were not randomly assigned to interventions, and there may have been variation in how providers selected regimens in collaboration with families. We did not have data on children’s insurance status, and cannot evaluate the impact that needing to purchase medication may have had on adherence. We assumed that families not receiving medication through health departments were nonadherent if they did not return for clinic visits. Although this may have led to underestimating completion rates, few of our patients have regular pediatricians who may have been able to continue therapy. Furthermore, we could not evaluate for AEs in children who did not return for follow-up, which may have resulted in underestimating AEs. It is possible that children could have moved to other jurisdictions and then developed TB disease; thus, the effectiveness of therapy could have been overestimated; for this reason, effectiveness was an exploratory outcome. The results of our findings may not be generalizable to other clinical settings, particularly those in which 3HP or health department assistance with medication administration may not be readily available.
We demonstrated that short-course regimens were associated with significantly higher completion rates compared with 9H and were well tolerated. The effectiveness of these regimens is high in the short-term. We hope that the coming years will see a shift in TB infection diagnosis to IGRAs to enable treatment to target the children who would receive the most benefit from therapy, and that increased data on 3HP pharmacokinetics and more child-friendly formulations of 3HP will enable this shorter regimen to be used in the most vulnerable pediatric population: infants and young children.
directly observed preventive therapy
enhanced self-administered therapy
interferon γ release assay
liver function test
tuberculin skin test
3 months of once-weekly isoniazid and rifapentine
4 months of daily rifampin
9 months of daily isoniazid
Dr Cruz conceptualized and designed the study, performed data collection and analyses, drafted the initial manuscript, and reviewed and revised the manuscript; Dr Starke helped conceptualize the study and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
FINANCIAL DISCLOSURE: Dr Starke is on the data safety monitoring board for Otsuka Pharmaceuticals for pediatric pharmacokinetic studies of delamanid; and Dr Cruz has indicated she has no financial relationships relevant to this article to disclose.