INTRODUCTION
Tuberculosis (TB) in older people (aged ⩾65 years) is a significant public health problem in low TB-incidence countries, especially in nursing homes and long-term care facilities [Reference Thrupp1–Reference Rajagopalan and Yoshikawa6]. Ageing is accompanied with TB risk factors; malnutrition, sarcopenia, progressive reduction in cell-mediated immune function and co-existing illnesses such as diabetes mellitus, chronic renal failure, malignancy and antineoplastic chemotherapy [Reference Thrupp1–Reference Yoshikawa4]. Elderly persons have increased incidence of TB, higher TB reactivation and higher TB mortality. Active TB at the time of arrival at the nursing home, reactivation of latent tuberculosis infection (LTBI) and transmission within the nursing home are causes of TB in the elderly. The low treatment completion proportion and high risk of adverse reaction due to LTBI treatment are also seen in the elderly population [Reference Pratt3, Reference Yoshikawa4]. A delay in diagnosis and initiation of TB treatment with atypical clinical and radiological features is a significant factor for widespread transmission in nursing homes and long-term care facilities [Reference Rajagopalan5]. A TB outbreak needs large-scale contact screening to detect LTBI and to control TB in nursing-home residents, healthcare workers, their staff and persons in the local community after prolonged TB exposure [Reference Thrupp1, Reference Hochberg and Horsburgh2, Reference Rajagopalan and Yoshikawa6].
Two Mycobacterium tuberculosis-specific interferon-gamma release assays (IGRAs), QuantiFERON®-TB Gold In-Tube (QFT; Qiagen, Germany) and T-SPOT®.TB (T-SPOT; Oxford Immunotec, UK), are available instead of the tuberculin skin test (TST) as new methods for diagnosing LTBI. IGRAs neither influence by bacillus Calmette-Guérin (BCG) vaccination nor have booster phenomenon, unlike TST. They have excellent accuracy with higher sensitivities and specificities than those of TST, especially in BCG-vaccinated individuals [Reference Pai, Zwerling and Menzies7–Reference Diel9]. However, the purchase cost of IGRAs are higher than those of TST and chest X-ray (CXR) examination.
The global proportion of older people increased from 9·2% in 1990 to 11·7% in 2013 and will continue to grow as a proportion of the world's population, reaching 21·1% by 2050 [10]. Cost-effectiveness regarding the use of IGRAs for TB screening for nursing homes warrants evaluation as a TB policy control measure.
In this study, cost-effectiveness of TB screening using IGRA (QFT or T-SPOT), compared to TST, TST followed by QFT or T-SPOT, CXR for active TB screening, and no screening was assessed to evaluate the optimal entry method for older persons to nursing homes.
METHODS
Target population
The target population was a hypothetical cohort of 84-year-old residents and those with comorbidities such as HIV infection, diabetes mellitus and chronic kidney disease in nursing homes using a societal perspective on a lifetime horizon. Nursing homes are defined as institutions that provide healthcare to people who are unable to manage independently in the community. The average age at nursing-home entry is 84 years and the average time spent living there is 4 years [11]. In Japan almost all the elderly have received BCG vaccination.
As this was a modelling study with all inputs and parameters derived from published literature, ethical approval was not required.
Decision trees and Markov models
The following seven clinical states were included in this model to represent the possible clinical states in the target populations: (i) well (no LTBI, no TB); (ii) LTBI; (iii) LTBI, taking LTBI treatment without complication; (iv) LTBI, taking LTBI treatment with liver dysfunction; (v) drug-sensitive TB (DS-TB) during TB treatment and before; (vi) multidrug resistant tuberculosis (MDR-TB) during MDR-TB treatment and before; (vii) dead. Decision-analytical calculations were performed using TreeAge Pro Healthcare Module 2012 (TreeAge Software Inc., USA). Each cycle length was 1 year.
Decision trees and Markov models were developed for seven strategies; no screening, TST, QFT, T-SPOT, TST followed by QFT, TST followed by T-SPOT, and CXR (Fig. 1). Per-person cost and effectiveness were calculated. The incremental cost effectiveness ratio (ICER) of each screening arm was applied and compared. The rates of adherence of chemoprophylaxis, liver dysfunction induced by chemoprophylaxis, and completion of DS-TB and MDR-TB treatments were considered. Markov models that took into account comorbidities such as HIV infection, diabetes mellitus and chronic kidney disease were also constructed with the lower test sensitivities and their relative risks of reactivation rates.
-
(1) No screening
-
(2) TST strategy. A nursing-home resident undergoes TST testing. If TST induration diameter is ⩾5 mm in those without BCG vaccination and ⩾10 mm in those with BCG vaccination, the resident undergoes CXR. If active TB is suspected based on CXR, and subsequent smears, cultures and drug sensitivity test of sputum examination are performed, the resident is treated with the standard 6-month protocol for DS-TB or the protocol for MDR-TB. If active TB is not detected by CXR, the resident receives 9-month isoniazid (INH) chemoprophylaxis. If TST induration diameter is <5 mm in those without BCG vaccination and <10 mm in those with BCG vaccination, the resident does not require follow-up. The proportion of residents for whom the TST was performed and read was 1·0.
-
(3) IGRA (QFT or T-SPOT) strategy. A nursing-home resident undergoes IGRA testing. If the IGRA is positive, active TB is suspected based on CXR and subsequent smears, and cultures and drug sensitivity test of sputum examination are performed, the resident is treated with the standard 6-month protocol for DS-TB or the protocol for MDR-TB. If the IGRA is positive and active TB is not detected by CXR, the resident receives 9-month INH chemoprophylaxis. If the IGRA is negative, the resident does not require follow-up.
-
(4) TST followed by IGRA (QFT or T-SPOT) strategy. A nursing-home resident undergoes TST testing. If TST induration diameter is ⩾5 mm in those without BCG vaccination and ⩾10 mm in those with BCG vaccination, the resident undergoes IGRA testing. If the IGRA is positive, active TB is suspected based on CXR and subsequent smears, and cultures and drug sensitivity test of sputum examination are performed, the resident is treated with the standard 6-month protocol for DS-TB or the protocol for MDR-TB. If the IGRA is positive and active TB is not detected by CXR, the resident receives 9-month INH chemoprophylaxis. If the IGRA is negative, the resident does not require follow-up. If TST induration diameter is <5 mm in those without BCG vaccination and <10 mm in those with BCG vaccination, the resident does not require follow-up.
-
(5) CXR strategy. A nursing-home resident undergoes a CXR test. If CXR is positive, active TB is suspected based on CXR and subsequent smears, cultures and drug sensitivity test of sputum examination are performed, the resident is treated with the standard 6-month protocol for DS-TB or the protocol for MDR-TB. If CXR is negative, the resident does not require follow-up.
Fig. 1. Simplified illustration of the decision trees. A square node represents the decision node. A circular node represents a chance node. Branches from a chance node represent possible outcomes. An M○ node represents a Markov node. QFT, QuantiFERON®-TB Gold In-Tube; TB, tuberculosis; T-SPOT, T-SPOT®.TB; TST, tuberculin skin test; CXR, chest X-ray examination; INH, 9-month INH chemoprophylaxis protocol for latent tuberculosis infection.
Probabilities, costs, effectiveness, utilities and other assumptions
All data were collected using Medline. A search of the literature published from 1980 to 2 April 2016 was undertaken to use incremental cost-effectiveness analysis.
Prevalence of LTBI and TB, probability of TB patients having MDR-TB, relative risk of TB in the elderly, adherence rate of chemoprophylaxis, probability of hepatotoxicity induced by chemoprophylaxis, efficacy of chemoprophylaxis protocol, the completion rates of DS-TB and MDR-TB treatments, recurrence rates of DS-TB and MDR-TB after treatment and mortality rates of DS-TB and MDR-TB were derived from the published literature [Reference Hochberg and Horsburgh2, Reference Fountain12–Reference Horsburgh23]. The BCG vaccination rate was 0·93 in Japan in 2012 [24]. Age-specific all-cause mortality rates were obtained from Japanese life tables. Data from the meta-analyses, which included studies from numerous low-incidence countries, were used to determine the sensitivities and specificities of TST, QFT and T-SPOT [Reference Pai, Zwerling and Menzies7–Reference Diel9, Reference Kowada26–Reference Faurholt-Jepsen29]. The sensitivity and specificity of CXRs were obtained from the published literature [Reference Tattevin30]. The lower test sensitivities and the relative risks of reactivation rates in the elderly with comorbidities such as HIV infection, diabetes mellitus and chronic kidney disease were also obtained from the published literature [Reference Horsburgh23, Reference Kowada26–Reference Faurholt-Jepsen29].
Cost data were collected using a societal perspective. All costs were adjusted to 2014 Japanese yen, using the medical care component of the Japanese consumer price index and were converted to US dollars (US$), using the Organisation for Economic Cooperation and Development (OECD) purchasing power parity rate in 2014 (1 US$ = ¥105·8) [31, 32]. The cost of TST screening included labour costs for two physician visits and the TST reagents. The costs of QFT and T-SPOT screening included the screening kits, one physician visit, and the labour costs for laboratory technicians [31, 33]. The cost of CXR screening included the material cost of CXR, one physician visit, and the labour costs for radiology technicians [31, 33]. The costs of TB treatment, 9-month INH chemoprophylaxis and treatment of liver dysfunction caused by chemoprophylaxis were determined from the national fee schedule in Japan [31] (Table 1). The costs of smears, cultures and drug sensitivity testing of sputum examinations were also considered [31]. All costs were discounted at a fixed annual rate of 3%. Per-person costs were calculated for each strategy.
Table 1. Baseline estimates for selected variables
* 95% confidence interval.
BCG, Bacillus Calmette-Guérin; CXR, chest X-ray examination; DS-TB, drug-sensitive tuberculosis; IGRA, interferon-gamma release assay; LTBI, latent tuberculosis infection; MDR-TB, multidrug-resistant tuberculosis; QFT, QuantiFERON®-TB Gold In-Tube; TB, tuberculosis; T-SPOT, T-SPOT®.TB; TST, tuberculin skin test; 9H, 9-month isoniazid.
The main outcome measure of effectiveness was quality-adjusted life-years (QALYs). Health state utilities were calculated by using a utility weight of 0·58 for MDR-TB, 0·80 for DS-TB, 0·85 for LTBI (taking chemoprophylaxis with complication), 0·95 for LTBI (taking chemoprophylaxis without complication) and 1 each for LTBI and well (Table 1) [Reference Guo, Marra and Marra34, Reference Dion35]. All clinical benefits were discounted at a fixed annual rate of 3%. Per-person QALYs were calculated for each strategy.
One-way sensitivity analyses and probability sensitivity analyses
One-way sensitivity analyses and probability sensitivity analyses were performed to determine which strategy yielded the greatest benefits and costs, using the ranges of probabilities, costs, relative risks and utilities. Each model variable was assigned a distribution based on the values in the literature and assumptions (Table 1). By Monte Carlo simulation distributions, the selected probabilities are in β distributions and the selected relative risks are in lognormal probabilities.
RESULTS
In the base-case analysis, QFT strategy was the most cost-effective at the willingness-to-pay level of US$ 50 000/QALY gained (US$ 401·9, 4·36 707 QALY; ICER 91·3 US$/QALY, year 2014 values). TST followed by QFT strategy (US$ 516·3, 4·36 900 QALY; ICER 59 129·9 US$/QALY) was less cost-effective than QFT strategy. CXR strategy (US$ 6683·3, 4·37 579 QALY; ICER 908 961·6 US$/QALY) was less cost-effective (Table 2). In analyses considered with higher risk of TB reactivation due to comorbidities such as HIV infection, diabetes mellitus and chronic kidney disease, TST followed by QFT strategy was more cost-effective than QFT strategy. CXR strategy was also less cost-effective (Table 2).
Table 2. Results of seven strategies for TB screening of elderly nursing-home residents
CXR, Chest X-ray examination; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-year; QFT, QuantiFERON®-TB Gold In-Tube; T-SPOT, T-SPOT®.TB; TST, tuberculin skin test.
One-way sensitivity analyses
In the base-case analysis, cost-effectiveness was sensitive to LTBI rate and BCG vaccination rate. TST followed by QFT strategy was more cost-effective than QFT strategy when the LTBI rate was >0·35 and when the BCG vaccination rate was <0·57 at the willingness-to-pay level of US$ 50 000/QALY gained (Tables 3, 4). In the analyses considered with the risk of TB reactivation due to comorbidities such as HIV infection, diabetes mellitus and chronic kidney disease, QFT strategy was more cost-effective than TST followed by QFT strategy when the LTBI rate was <0·18 in HIV-infected residents, 0·30 in diabetes mellitus residents, 0·26 in chronic kidney disease residents, and when the BCG vaccination rate was >0·95 in diabetes mellitus residents at the willingness-to-pay level of US$ 50 000/QALY gained.
Table 3. Sensitivity analysis of LTBI rate
CXR, Chest X-ray examination; ICER, incremental cost-effectiveness ratio; LTBI, latent tuberculosis infection; QALY, quality-adjusted life-year; QFT, QuantiFERON®-TB Gold In-Tube; T-SPOT, T-SPOT®.TB; TST, tuberculin skin test.
Table 4. Sensitivity analysis of BCG vaccination rate
BCG, Bacillus Calmette-Guérin; CXR, chest X-ray examination; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-year; QFT, QuantiFERON®-TB Gold In-Tube; T-SPOT, T-SPOT®.TB; TST, tuberculin skin test.
Hepatotoxicity by 9-month INH chemoprophylaxis had little impact on cost-effectiveness.
Probabilistic sensitivity analyses
According to the Monte Carlo simulations for 10 000 trials, the cost-effectiveness acceptability curve in 84-year-old nursing-home residents demonstrated that the QFT strategy had more chance of being cost-effective than TST followed by QFT with 65% probability at the US$ 50 000 willingness-to-pay level (Fig. 2).
Fig. 2. Cost-effectiveness acceptability curve. QFT, QuantiFERON®-TB Gold In-Tube strategy; TST/QFT, tuberculin skin test followed by QuantiFERON®-TB Gold In-Tube strategy.
DISCUSSION
This study demonstrated that using IGRA was more cost-effective for LTBI screening for nursing homes at the willingness-to-pay level of US$ 50 000/QALY gained. Highest specificity of QFT, and cost savings due to effective chemoprophylaxis by preventing TB reactivation in the elderly are the main reasons for the higher cost-effectiveness result of the QFT strategy. Cost-effectiveness was sensitive to LTBI rate and BCG vaccination rate. Hepatotoxicity by 9-month INH chemoprophylaxis had little impact on cost-effectiveness.
In this study, the main outcome measure of effectiveness was QALYs gained. The use of QALYs can combine the effects of quantity of life with quality of life in a single measure. ICER, which is calculated by using incremental costs and incremental QALYs gained can be compared to the willingness-to-pay level. Willingness to pay provides a measure of the societal value attached to a given health benefit when the values from a population are aggregated. Even if the differences in effectiveness between IGRA and TST on LTBI screening are very small, as in this case, the willingness-to-pay method using QALYs gained is very useful to evaluate cost-effectiveness.
Current entry TB screening is conducted by CXR as active TB screening in Japan. There is no data regarding the prevalence of hepatotoxicity by 9-month INH chemoprophylaxis in nursing-home residents in Japan. We derived the prevalence of hepatotoxicity and efficacy of the chemoprophylaxis protocol from the published literature. CXR examination only can detect active TB. When TB is detected by CXR screening for symptoms of TB in nursing-home residents, TB infection spreads in nursing residents and healthcare staff [Reference Ijaz36, Reference Chitnis37]. Large-scale contact screening in nursing homes is needed [Reference Ijaz36, Reference Chitnis37]. Some residents may die due to transmission of TB in nursing homes [Reference Rajagopalan5, Reference Rajagopalan and Yoshikawa6, Reference Ijaz36, Reference Chitnis37]. This study demonstrates that active case-finding was not cost-effective and that preventive strategy with the diagnosis and treatment of LTBI was the most efficient strategy to control TB in nursing homes despite its hepatotoxicity in low TB-incidence countries.
A previous study reported the cost-effectiveness of QFT compared to CXR, and no screening compared to TB screening of the BCG-vaccinated elderly general population and demonstrated that the no-screening strategy offered the greatest cost saving for elderly populations in Japan [Reference Kowada38]. In that study for the elderly general population, the results also demonstrated that the QFT strategy was more cost-effective than no screening when TB prevalence was >0·00 047 on the sensitivity analysis. The superiority of the QFT strategy in the present study is consistent with those results.
To the best of our knowledge, this study is the first cost-effectiveness analysis of IGRAs for TB screening of elderly nursing-home residents, compared to TST, TST followed by IGRAs, CXR and no screening using a Markov model.
Katsenos et al. showed that QFT had a significant additive value to single TST for detecting LTBI in institutionalized older adults [Reference Katsenos20]. Verma et al. demonstrated that LTBI screening with TST for the elderly was more cost-effective than CXR screening in long-term care facilities in Canada and concluded that TB screening strategies on entry to long-term care are costly [Reference Verma, Chuck and Jacobs18]. We first demonstrated that using IGRA was more cost-effective than TST and CXR for entry TB screening to a nursing home.
There are several limitations to this study. First, sensitivities and specificities of TB screening kits (IGRA and TST), were obtained from meta-analyses of immunocompetent individuals, but not for older people with waning immunity. Further study of test sensitivities with waning immunity of the elderly is needed. Second, there is little data on LTBI rates using IGRAs in nursing-home residents. Further studies of the elderly based on IGRA testing are needed. Third, the harm from radiation exposure by repeating CXR was not considered in this model. Fourth, the use of rifapentine plus isoniazid for 3 months, which had a higher treatment completion rate, was not considered for chemoprophylaxis regimen of nursing-home residents in this model. Further, long-term safety monitoring research and a cost-effectiveness study using rifapentine plus isoniazid for 3 months is required. Fifth, the epidemiology of TB in the elderly needs to be dealt with in much more detail in order to make a more convincing case for TB policy change. Sixth, there is no method for diagnosing whether LTBI differentiates first infection with TB from reinfection. Seventh, there are little epidemiological studies of TB outbreaks in nursing homes. Finally, there are different costs and medical systems in each country. Further cost-effectiveness studies will be needed for each country using each cost.
CONCLUSIONS
QFT [US$ 401·9, 4·36 707 QALY (year 2014 values)] was the most cost-effective at the willingness-to-pay level of US$ 50 000/QALY gained. TST followed by QFT was the most cost-effective in residents with comorbidities. CXR was less cost-effective. Effective LTBI screening using IGRA is recommended to prevent TB transmission not only in nursing homes but also in local communities in low-incidence countries.
DECLARATION OF INTEREST
None.
