Health Technology Assessment Of Two Antituberculosis Regimens And A Correspondence Analysis Between Tuberculosis Inpatient Attributes And Treatment Results Obtained In General Hospitals Of China

Background:Tuberculosis (TB) is a health problem that is on the increase worldwide.China, with the latest estimated TB prevalence rate of 246/10~5 and TBincidence rate of 102/10~5, is one of 22 high-burden countries (HBCs). Thenumber of TB cases per year in China rank the second in the world, includingabout 17% of all new TB cases. The surveillance data from ministry of healthof China showed that tuberculosis replaced hepatitis B as the leadingcontagion of China in 2005 Mar.Chemotherapy and chemoprophylaxis are mainstays of modem TBcontrol, and antituberculosis drug is their base. It is nesscery to assess theefficacy, safety and cost-effectiveness of antituberculosis-drug or regimensbecause duration of antituberculosis-drug required at least 6 months generallybrings more adverse effects. If antibuerculois-drug with bad clinical edffectsis ingested by patients, treatment effect expected may not be achieved, whilebody injure, multidrug-resistant tuberculosis and more costs may be produced.This study assessed the efficacy,safety and cost-effectiveness of tworifamycin-based regimens (short-course rifampicin plus pyrazinamide, containing rifapentine regimen) by using systematic review of evidence.The directly observed treatment, short course (DOTS) strategy, as themost important health-intervention method formed in the early 1990s, is themost effective strategy for controlling tuberculosis. To address the threat ofTB, the National TB Control Program (NTP) was initiated in the early 1990s,mainly with funding from a World Bank loan and the Ministry of Health,China. In NTP-covered areas, DOTs strategy recommended by the WorldHealth Organization is followed. Although the geographical coverage ofDOTS increased substantially between 2003 and 2005(from 91% in 2003 to100% in 2005), there is a long way for China from the case detection rate of45% in 2003 to reach the 70% global target. The year 2000 prevalence surveyreported that detecting TB case in China mainly depended on patients' accessto hospital (94.3%). But in China, system of hospitals still separates from thenational TB control program system, most of TB patients in China werediagnosed and treated in hospitals, especially in general hospitals, but only aminority were referred to TB dispensaries. Few studies have been conductedin China on treatment result of TB inpatients treated in hospitals, still lessgeneral hospitals. The question is whether the current hospitals in China canprovide a satisfactory curative effect for TB inpatients. The another aim ofthis study is to learn the actual treatment results of TB inpatients treated ingeneral hospitals by using database of TB inpatient discharged from generalhospitals of Nanchong region in 2003, further to explore and view the furtherrelationship between treatment results and different inpatients attributes bycorrespondence analysis, which is helpful to identify those subgroups of TBpatients for which intervention is a priority.Method:1. By using quantitive systematic review of evidence, we assessed theclinical effects and safety of the short-course rifampicin plus pyrazinamide regimen(RZ) versus the standard isoniazid regimen (H)for the treatment oflatent tuberculosis infection (LTBI), and that of the containing rifapentineregimen versus the comparative containg-rifampicin regimen for thetreatment of tuberculosis,respectively.We included randomized controlled trials (RCTs) and quasi-RCTs thatcompared the regimen of rifampicin plus pyrazinamide given for 2-3 monthswith the regimen of isoniazid given for 6-12 months for treatment of LTBIfor assessing the short-course RZ regimen, and included randomizedcontrolled trials that compared rifapentine-based regimen with comparativerifampicin-based regimen for treatment of tuberculosis for assessingrifapentine regimen. Considering language barrier, only trials expressed inEnglish or Chinese language were included in the study. We searchedpublished information in Medline (1996 to Mar 2007), Embase (1984 to2004), the Cochrane Central Register of Controlled Trials (CENTRAL,Cochrane Library issue 1,2007), BIOSIS preview (1997 to 2006), ChineseBiomedical Literature Database (CBM, up to 2006), and Vip database(Chinese, 1989 to 2006), as well as unpublished information in Wanfangdatabase (Chinese, up to 2006) and ClinicalTrials.gov. Search words of"tuberculosis","rifampicin","rifampin","pyrazinamide","rimifon","isoniazid","prophylaxis","chemoprophylaxis","prevention andcontrol" and "preventive therapy" were used for assessing the short-courseRZ regimen, and search words of "tuberculosis" and "rifapentine" were usedfor assessing the rifapentine regimen,respectively.We used checklists to assess the methodology quality of the includedtrials. For assessing the efficacy and safety of two regimens, treatment effectswere summarized as risk difference (R.D) with 95% confidence interval (CI).Risk difference values of zero denoted no difference for the outcome. Z-testwas used for the pooled risk difference and a low P-value of＜0.05 was considered to be statistically significant. The heterogeneity of treatmenteffects between studies was investigated using chi-squared test and I~2statistics. A low P-value of≤0.1 was considered to be statisticalheterogeneity among studies. 'I~2' denotes the percentage of total variationacross the studies that is the result of heterogeneity rather than chance (an I~2value of 0% denotes no observed heterogeneity, and I~2 value of＞50% denotesobvious heterogeneity). If there was no statistical heterogeneity amongstudies, pooled risk difference of treatment effects was calculated using afixed effect model. If there was statistical heterogeneity among studies,subgroup analysis and sensitivity analysis were planned to explore sources ofheterogeneity, including trial quality, treatment duration, and drug dosage. Weassessed for the presence of publication bias by use of a funnel plot. Ifsources of heterogeneity couldn't be found, pooled risk difference oftreatment effects was calculated using random effect model.Qualitative systematic review was used to assess the cost-effectivenessof two regimens by identifying economic studies from these articles searchedfor assessing the efficacy and safety of two regimens respectively.2. A retrospective study was designed for inpatients with tuberculosisdischarged from general hospitals of Nanchong region in Sichuan province in2003. Descriptive analyses were performed in order to obtain a wholeimpression. Median was selected to stand for the average length of stay andinpatient costs because they are not normally distributed. A series ofchi-square tests were used to obtain statistically significant factors associatedwith treatment results and with TB categories respectively. A P-value of≤0.05 was considered as statistically significant. Correspondence analyses(CA) were undertaken to further visualize the relations between differentlevels of these factors, different TB categories and different treatment results. Results:1. short-course rifampicin plus pyrazinamide versus isoniazid fortreating latent tuberculosis infection.To assess the efficacy and safety of this regimen, 5 randomizedcontrolled trials (RCTs) and 1 quasi-RCT were identified. Of the 6 trials, onewas rated high-quality, four were low-quality, and the remaining one wasmoderate-quality. 3 trials were conducted in HIV-infected patients and theother 3 trials were conducted in HIV-uninfected persons. The rates oftuberculosis in the rifampicin plus pyrazinamide group were similar to that inthe isoniazid group, whether the subjects were HIV-infected patients or not(for HIV-infected patients: pooled RD=0%, 95%CI:-1% to 2%, P=0.89; forHIV-uninfected persons: pooled RD=0%, 95%CI:-2% to 1%, P=0.55). Therewas no difference in mortality between the two treatment groups (forHIV-infected patients: pooled RD=-1%, 95%CI:-4% to 2%, P=0.53; forHIV-uninfected persons: pooled RD=0%, 95%CI:-1% to 1%,P=1.00).whether the RZ regimen would increase the risk of sever adverseevents in HIV-infected persons compared with the isoniazid regimen was notdetermined clearly in this study.One subgroup analysis involving 2 trials showed no significantdifference in severe hepatotoxicity or all severe adverse events betweenHIV-infected patients who received RZ regimen for 2-3 months(twice-weekly) and those who received isoniazid for 6 months (twice-weekly)(severe hepatotoxicity: RD=0%,95%CI:-1% to 0%,P=0.37; all severe adverseevents:RD=0%,95%CI:-1% to 2%,P=0.69). However, another subgroupanalysis including one trial showed that higher incidence of all severe adverseevents but lower risk of severe hepatotoxicity was associated with 2RZregimen daily than 12H regimen daily.(severe hepatotoxicity: RD=-2%,95%CI:-3% to 0%,P=0.01; all severe adverse events:RD=4%,95%CI:1% to 7%,P=0.004).For HIV-uninfected persons, both subgroup analyses showedthat a higher incidence of all severe adverse events was associated withrifampicin plus pyrazinamide than isoniazid (one: RD=29%, 95%CI: 13% to46%; P=0.0005; another: RD=7%, 95%CI:4% to 10%; P＜0.0001).To assess the cost-effectiveness of this regimen, 3 studies with highquality were identified. The cost of the 12H regimen daily per QALY savedwas the cheapest for treating HIV-infected persons; the cost of 2RZ regimentwice-weekly per QALLY saved was lower than the cost of 6H regimentwice-weekly per QALY saved for the treatment of HIV-infected LTBI. Thecost of 2RZ regimen per life expectancy saved was higher than that of the 6Hregimen per life expectancy saved for the treatment of HIV-uninfected LTBI.2. Rifapentine-based regimen versus rifampicin-based regimen fortreatment of tuberculosis.To assess the efficacy and safety of this regimen, 11 randomizedcontrolled trials (RCTs) were identified. Of the 11 trials, 2 trials were ratedhigh-quality, 5 trials were low-quality, and the remaining 4 trials weremoderate-quality. 11 trials included in the study were all conducted inpulmonary tuberculosis patients, of which, 1 trial was performed in treatingsmear-negative and culture-negative pulmonary tuberculosis patients(bacteriology-negative TB) and 10 trials were performed in treatingsmear-positive or culture-positive pulmonary tuberculosis patients(bacteriology-positive TB). For treating initial bacteriology-positive TB, allsubgroup analyses showed that the pooled risk difference with 95%CI ofsputum negative conversion rate at completion of full course, evere adverseeffects rate, and of severe hepatotoxicity rate all included "0", P＞0.05, butthe subgroup analyses involving once-weekly and less frequency rifapentinetherapy respectively compared with intermittent rifampicin therapy showedthat the pooled risk difference of bacteriological relapse rate were 3% (95%CI:1% to 5%, P=0.06), and 5% (95%CI:1% to 9%, P=0.01)respectively. There was statistically significant differene between the twogroups for this outcome. No economic studies were identified to assess therifapentine-based regimen.3. Correspondence analysis between TB inpatient attributes andtreatment results obtained in general hospitals.Statistically significant factors associated with TB categories weregender, age, and marital status. Statistically significant factors associatedwith treatment results were TB categories, gender, age, comorbidity, lengthof stay (LOS), and illness state at access to hospital.The CA showed mainresults as follows:(1) TB for inpatients, who were male, married, aged≥15 yrs, morelikely happened in respiratory organs, and that for female inpatients morelikely happened in other organs. TB treatment result of hospitalization wasmore likely to be "improved" for inpatients with TB happened in respiratoryorgans. (2) Attributes of inpatients died in general hospitals were clearlydifferent from that of cured and improved inpatients. (3) TB treatment resultof hospitalization was more likely to be "improved" for TB inpatients whowere male, aged≥15 yrs, and with urgent illness state at access to hospitals,while TB Inpatients younger than 15 yrs tend to leave hospitals withoutdoctor's permission. (4) Treatment result of cure was relatively likely forfemale TB inpatients with LOS＞8 days, average illness state at access tohospitals, but without comorbidity. (5)TB inpatients with bad treatmentresults appeared some similarities of age＜15 yrs and dangerous illness stateat access to hospital.Conclusions:1. For treatment of HIV-infected LTBI, the cost of 12H regimen dailyper QALY saved is the cheapest, while the cost of 2RZ regimen twice weekly per QALY saved is lower than that of 6H regimen twice weekly. Fortreatment of HIV-uninfected LTBI, the cost of 6H regimen daily per lifeexpectancy is lower than that of 2RZ regimen daily. Considering efficacy andsafety of these regimens, the 2RZ regimen twice weekly is better than the 6Hregimen twice weekly in treating the HIV-infected LTBI.Whether the 2RZregimen daily is better than the 12H regimen daily in HIV-infected personscan not be determined because the 2RZ regimen daily is more expensive butless hepatotoxicity than 12H regimen daily. However, the 2RZ regimen dailyincreases risk of severe adverse effects compared with 6H regimen daily inHIV-uninfected persons, and its ratio of cost/effectiveness is higher than thatof 6H regimen daily. Not the 2RZ regimen daily but 6H regimen daily shouldbe recommended for treatment of HIV-uninfected LTBI.2. For treating initial bacteriology-positive TB, the efficacy and safety ofonce-weekly and twice-weekly rifapentine therapy is similar to that of dailyrifampicin therapy.Though the once-weekly and less frequency rifapentinetherapy and intermittent rifampicin therapy (twice-weekly or thrice-weekly)have similar short-term efficacy and safety, the once-weekly and lessfrequency rifapentine therapy increase the risk of bacteriological relapsecompared with intermittent rifampicin therapy.3. Due to the relationships between treatment results gained in generalhospitals and TB inpatients' attributes, a series of corresponding strategies forTB inpatients with special attributes should be made out to strengthencompletion treatment of TB and obtain higher cure rate.