Introduction

One-quarter of the world has latent tuberculosis infection (LTBI); an estimated 10% will eventually develop active tuberculosis (TB) [1]. This LTBI population functions as the reservoir driving ongoing incidence of active TB, even in the absence of continued transmission. Finding and treating LTBI is important to control TB and for the ultimate goal of eliminating TB [1,2,3].

For many years, LTBI has been treated with isoniazid administered daily for 9 months (9H), which reduces reactivation by 90% but incurs risk of hepatoxicity [4, 5]. Isoniazid is widely-available, and costs very little.

More recently, shorter options for LTBI treatment have been adopted [6], including a three-month regimen of once-weekly isoniazid and rifapentine (3HP) [7]; 4 months of daily rifampicin (4R) [8]; or 3 months of daily isoniazid and rifampicin [6]. Shorter courses of treatment are similarly effective and easier for patients to complete [9]. 4R is associated with a theoretical risk of rifampicin resistance, though not empirically demonstrated [10]. 3HP is effective [7] but is usually given via directly-observed treatment (DOT), which requires more visits and human resources, and has corresponding budgetary implications. Self-administration of 3HP met the threshold for non-inferiority to DOT in [11], and is an approved strategy for, the US [12]. However, rifapentine is significantly more expensive than isoniazid.

Directly-observed preventative therapy (DOPT) for treatment of LTBI is a commonly used programmatic strategy, particularly in populations at high-risk of developing active TB [13,14,15,16,17]. This treatment support might mitigate the risk related to congregate or crowded living [14, 15], cultural and/or linguistic barriers [13, 14]. Barriers such as lack of trust in authorities, and histories of racism and colonialism may pose challenges for DOPT [18, 19]. While DOPT can be used with any LTBI regimen, it is recommended with 3HP because of the weekly dosing interval and significant pharmacokinetic impact of a missed dose.

A 2011 systematic review evaluated cost effectiveness evidence for LTBI treatment overall [20]. It excluded high-risk populations including HIV-coinfected patients, but did not restrict drug regimens, including self-administered regimens and DOT. Isoniazid was cost-effective for the general population in high-resource settings compared to no treatment, but there was insufficient evidence for low-resource settings. The systematic review included only one study directly comparing two drug regimens.

Since 2011, several more cost-effectiveness studies have been published. With 3HP, the requirement for direct observation coupled with expensive medication incurs more up-front costs and could be cause for policymakers to hesitate. Economic evidence to inform these decisions is critical.

We performed a systematic review on the cost-effectiveness of 3HP, recognizing potential variation across settings, geographic regions, and specific populations.

Methods

A systematic review was performed to determine the cost and cost-effectiveness of 3HP DOT. We searched PubMed, Embase, CINAHL, LILACS, and Web of Science up to February 28, 2022, with search concepts combining latent tuberculosis, drug therapy or directly observed preventive therapy, and cost, economic, or cost effectiveness (see Additional file 1 for full details on the search strategy).

Studies included were in English or French, on human subjects, with LTBI, provided information on specified anti-tubercular regimens, had a directly-observed arm, and described outcomes with some cost or economic data. We excluded posters and abstracts, studies on drug-resistant TB, and studies that combine testing and treating as a single intervention. Two reviewers (KB, WAL) screened all records independently; disagreements were resolved by consensus and if necessary by a third reviewer (AAZ). We included only studies examining 3HP DOT cost or cost-effectiveness. Data extraction was based on modified criteria from CHEERS checklist [21]. The eligible studies were extracted independently by two reviewers (KB, WAL).

Where there was no description of whether treatment was self-administered therapy (SAT) or administered by DOT, we assumed the usual strategy of administration based on drug regimen (e.g. 3HP was originally studied as DOT [7], 9H is usually self-administered) or based on national standards (e.g. LTBI treatment in Taiwan is routinely given by DOT) or by inference from information given regarding the frequency of visits or dosing.

We included studies that report data on cost with no information or follow-up with regard to effectiveness. We report on cost studies and cost-effectiveness studies separately, with outcomes on cost per patient and incremental cost effectiveness ratios (ICERs), respectively. For cases when authors provided itemized costs, such as for medications, medical professionals’ time, and laboratory and radiology investigations, we calculated estimated total costs per regimen for comparison. Quality of the included studies for the cost and cost-effectiveness review was assessed independently by two reviewers (KB, WAL) based on a modified Drummond checklist [22]. We did not plan for a meta-analysis given the range of geographical and contextual factors, including baseline adherence rate, LTBI prevalence, co-morbidity prevalence particularly HIV co-infection, usual care for LTBI treatment, and patient expectations.

For cost-effectiveness studies, we present results on 3HP DOT and 9H SAT. For studies in which there was no 9H SAT arm, we compared 3HP DOT to the comparator most similar to 9H SAT for which data was available (e.g. 9H DOT). ICERs were re-calculated using 9H SAT as baseline where possible. Costs have been converted to 2020USD based on Consumer Price Index [23].

Cost and cost-effectiveness analysis studies on self-administered 3HP (3HP SAT) alone, with no DOT arm, were excluded from the formal systematic review. Our primary outcome is cost and cost-effectiveness for 3HP.

Results

We identified 1937 records from database searches, 730 were duplicates, leaving 1207 for screening. Of these, 165 underwent full text review. Ten papers were included in this systematic review (see Fig. 1). We identified three costing-only studies and seven cost-effectiveness studies. Key study parameters are shown in Table 1. (See Additional file 2 for full details.)

Fig. 1
figure 1

PRISMA 2009 Flow Diagram. Directly-observed 3HP cost and cost-effectiveness systematic review [24]

Full size image
Table 1 Study characteristics
Full size table

Costing-only studies

The three costing-only studies compare treatments that are directly observed; none has a SAT group. Two are from hospitals in Taiwan [25, 26], where DOPT is the standard of care. The other pertains to Californian prisoners [27]. All 3 costing studies reported completion rates, and consistently reported improved completion rates [25,26,27] for 3HP compared to 9H (See Table 2).

Table 2 Cost/patient, as found in costing studies and compared to input costs/patient in cost-effectiveness studies, adjusted to 2020 USD, and completion rates
Full size table

All three costing studies are in contexts that have systems in place for DOT, so no new investments had to be made. With shorter duration and longer dosing interval, 3HP is favoured because of fewer total visits.

Treatment completion rates vary widely between studies, particularly for 9H (see Table 2). This heterogeneity may reflect regional or cultural differences in treatment acceptance and adherence.

Costing studies did not include any downstream costs of LTBI treatment, notably for serious adverse events (SAEs) and for active tuberculosis. Therefore, they had fewer component costs compared to the cost-analysis studies (see Table 3).

Table 3 Items included in cost inputs for cost-effectiveness and cost studies
Full size table

Cost-effectiveness studies

Of seven identified cost-effectiveness studies, four are modelled on US data, a high-resource setting [28,29,30,31]. Pease [32] focuses on the Canadian Arctic that is high income, high TB prevalence, geographically remote, and a majority-Indigenous population. Johnson [33] and Ferguson [34] examine cost-effectiveness in a low-resource, high HIV prevalence setting, using a Ugandan HIV clinic as their model, with different reactivation rate, mortality rate, treatment standards, and willingness-to-pay (WTP) (Table 4 and Additional file 3) compared with North America.

Table 4 Selected epidemiologic and cost input parameters in 2020 USD used across included studies
Full size table

The five cost-effectiveness studies in high-income settings vary on whether 3HP or 9H is better tolerated (Table 4). Pease [32] is unique with less difference in treatment completion between 3HP and 9H DOT, with higher 9H treatment cost (since it was DOT rather than SAT); and lower 3HP DOT treatment cost. Additionally, the cost of treating active TB is higher, likely due to geographical remoteness.

Five cost-effectiveness studies in high-income settings demonstrate widely-variable ICERs ranging from 3HP being cost-saving compared to 9H, to an ICER of $5418 per QALY gained, though all found that 3HP DOT is cost-effective with ICER values well under a WTP of USD $50,000 per QALY gained. Methods for the five were comparable, with a few inclusion differences such as contact tracing, secondary transmission, travel times, patient costs, and radiology (Table 3). All seven cost-effectiveness studies list one or more clinical outcomes; active TB cases per 1000 patients is common to all, ranging from 9.1 to 37 per 1000 patients treated with 9H, and 3.9 to 38 per 1000 patients treated with 3HP (Table 5).

Table 5 Outcomes, adjusted to 2020 USD
Full size table

Pease [32] compares 9H DOT against 3HP DOT in a high-incidence, high-income, remote Arctic setting and finds that 3HP DOT is cost-saving in almost all scenarios. Shepardson in a US setting makes analyses from health system and societal perspectives [30], demonstrating 3HP DOT’s ICER of USD $1081 per QALY gained from a societal perspective (Table 5). After a rifapentine cost-reduction, Shepardson recalculated the ICER and found 3HP DOT to be cost-saving compared to 9H SAT from a societal perspective [35]. Note that Shepardson reports effectiveness as mean QALY loss, rather than QALY gained, though ICERs are calculated as incremental cost per QALY gained (see Table 5).

Holland 2009 also directly compares 9H DOT to 9H SAT and found it not to be cost-effective, with an ICER well over USD $50,000 per QALY gained [28].

The study by Johnson, in a low-resource, high-prevalence setting with high HIV co-morbidity omits many costs, such as for treating adverse events or for hospitalization, while HIV treatment costs are included. This leads to higher overall cost per DALY averted [33]. Completion rates for both regimens are lower than other studies, particularly for 3HP DOT (see Table 4). Risk reductions on both regimens are lower, contributing to a higher ICER. Risk of adverse events, including drug-drug interactions, were assumed to be equivalent. Johnson concludes that in a low-resource setting, ICER is above a WTP of USD$1000/DALY averted (Table 5).

Based on a similar model, Ferguson [34] performed a cost-effectiveness analysis for 1HP compared to 3HP. 1HP is administered daily by self-administration, in contrast to 3HP which is administered weekly by DOT. There is no isoniazid arm and regimens are presumed equally effective [36], so this is not comparable to other studies and there is no inherent ICER. Compared to the Johnson study [33], input costs and overall costs are lower for 3HP DOT.

For high-resource settings, the ICER is expressed as dollars per Quality-Adjusted Life Year (QALY) gained but for a low-resource setting, the ICER is expressed as dollars per Disability-Adjusted Life Year (DALY) averted as per convention, and so these are not comparable.

The quality of the included studies, assessed on a modified Drummond checklist [22], is provided in Additional file 4. Of note, the costing studies [25,26,27] do not include all the costs incurred, particularly long-term, downstream costs/savings. They also do not include sensitivity analyses. Modelling studies in HICs by Doan [31], Holland [28, 29], and Shepardson [30] have no discussion related to ethical and distribution issues, though Pease [32] includes this. The modelling studies set in LMICs [33, 34] do not include the cost of adverse events. Chen [26] and Huang [25], who study Taiwanese patients, do not address generalizability of their findings.

Discussion and conclusion

This is the first systematic review on the cost-effectiveness of 3HP compared to 9H for LTBI. All studies in highly-resourced contexts found that 3HP DOT is cost-effective at a WTP of $50,000 per QALY gained, compared to 9H SAT or 9H DOT, despite the higher cost of rifapentine and the costs of DOT. Doan and Pease found it to be cost-saving [31, 32]. Shepardson’s 2014 update found 3HP to be cost-saving from a societal perspective [35]. However, in LMICs, 3HP DOT may not be cost effective. Johnson [33] found 3HP to have an ICER above a WTP of USD$1000/DALY averted. Ferguson did not compare 3HP to 9H [34].

Self-administered treatment studies were specifically excluded from this systematic review. Because missing a dose in a weekly regimen can lead to subtherapeutic drug levels, treatment failure, and development of drug resistance, we focus on 3HP DOT as it is habitually administered and studied. In a non-inferiority trial, 3HP SAT was found to be non-inferior to 3HP DOT in the US [11]. Scant literature examines 3HP SAT cost and cost-effectiveness. Denholm [37] demonstrated in a single Australian centre that 3HP SAT costs less than 9H SAT, at $375 compared to $441 USD per person treated, driven by more outpatient visits in the 9H arm. Yuen [38] examined costs of 3HP SAT compared to 6H SAT in Pakistan, demonstrating that in this LMIC, 3HP is also less costly, particularly after a rifapentine price reduction which resulted in 3HP SAT costing $294 USD compared to $399 for 6H SAT. Holland 2011 and Doan included a 3HP SAT regimen in their models: both studies found it more cost-effective than 3HP DOT and 9H SAT. [29, 31]

The heterogeneity found in the costing and cost-effectiveness studies might be in part because of qualitative aspects to how DOPT is operationalized. The literature supports measures that improve the ease of LTBI treatment, such as shorter courses of better-tolerated medication, such as 3HP and 4R in contrast to 9H [39, 40]. Providing treatment in locations and via structures that are convenient to patients also reduces barriers, for example in schools [41, 42], in residences [14, 15, 43], or with methadone treatment [16, 44,45,46]. The convenience of treatment is difficult to separate as a driver of adherence, from the effect of a DOPT strategy. DOT might also be a proxy for frequency of treatment support, as an opportunity for patients to ask questions, report side-effects, and generally engage with providers [47].

In contrast and depending on context and other program characteristics, DOT can also be perceived as punitive and paternalistic and therefore reduce trust and engagement with providers. Patients in high-risk populations find DOT to be humiliating and discriminating [18, 19]. Patients and providers alike acknowledge that the interaction to persuade compliance with DOT is based on authority and subtle threats [18]. Patients with positive experiences have opportunity to negotiate flexibility and had continuity of providers [18].

Our review has several limitations. The wide range of ICERs and of incident cases of active TB indicate uncertain findings. This is reinforced by variation regarding key sensitivity variables, such as expected increase in adherence rates. Several variables likely have local, regional, and cultural differences, for example the already-high baseline adherence rate in Taiwan. Cost of living and costs of medications and materials are higher in remote locations such the Canadian Arctic. Reactivation rates are highly correlated with HIV co-infectivity rates.

There are limitations in using conventional WTP measures. The ICER for 3HP DOT in a low-resource context [33] is above the conventional WTP—three times national GDP—but remains well below the cost of HIV treatment in that context, which is a baseline assumption of that model and an accepted, funded, real-world practice. This contradiction illustrates the utility of reporting relative cost-effectiveness in relation to other, accepted interventions [48]. Further, cost-effectiveness should be accompanied by other considerations in each context such as budget impact, feasibility, transparency, equity, and consistency [49].

From our systematic review, we conclude from the literature that 3HP DOT is cost-effective over the 9H SAT standard at WTP of less than $50,000 per QALY gained in high-income countries.