INTRODUCTION
Approximately one-third of the world's population is latently infected with tuberculosis (TB). In 2014, there were an estimated 9·6 million incident cases of TB, and 1·5 million people died from the disease [1].
In Taiwan, of all notifiable infectious diseases, TB has been the most prevalent for decades [2]. Since 2006, Taiwan's Centers for Disease Control and Prevention (CDC) has adopted a directly observed therapy short-course (DOTS) programme to halve TB incidence and achieve a successful treatment rate of 85% by 2015. Since then, the rate of TB incidence has decreased from 72·5 per 100 000 in 2005 to 48·4 in 2014 [2]. The success rate for TB treatment, however, only slightly improved from 64·1% in 2005 to 70·4% in 2014. Mortality among Taiwanese patients with TB accounted for 81·8% of the cases of unsuccessful TB treatment [2]. Additionally, over 82·1% of TB deaths in Taiwan were among individuals >65 years old [Reference Yen3].
TB in the elderly (>65 years old) has increasingly become a major issue not only because of non-specific clinical presentation of TB but, more importantly, because of a high mortality rate [Reference Cruz-Hervert4, Reference Byng-Maddick5]. Since the clinical symptoms of TB and risk factors of mortality in older patients may differ from those in younger patients, TB diseases in this population should be classified as a separate entity [Reference Morris6, Reference Schaaf7]. Few studies, however, have determined the factors associated with TB mortality among elderly patients [Reference Schaaf7].
According to the World Health Organization, death is defined as a patient who dies for any reason during treatment [Reference Balabanova8]. However, many patients with TB do not die of TB but from other causes such as malignancy or end-stage renal disease (ESRD). A previous review article found that few studies about TB treatment outcomes distinguished TB-specific mortality from other-cause (non-TB-specific) mortality [Reference Balabanova8]. Developing effective interventions to improve TB outcomes requires better understanding of the factors associated with mortality, particularly for vulnerable populations. We thus attempted to identify prognostic factors associated with mortality among elderly Taiwanese patients with TB infection from 2006 to 2014.
METHODS
Study population and data source
This retrospective cohort study used TB surveillance data collected by the Taipei City Government in Taiwan. The subjects included were elderly Taiwanese individuals (age ⩾65 years) with TB in Taipei from 2006 to 2014. In Taiwan, all suspected TB cases must be reported to Taiwan's CDC within 7 days [2]. After receiving notification, trained case managers use a structured questionnaire to interview patients about their socio-demographic characteristics, clinical findings, underlying diseases, admission history and TB treatments. Patients with TB in Taipei are required by law to be monitored until treatment success, death, or loss to follow-up. For the purpose of monitoring treatment response, case managers followed up all TB cases by phone or in person once every other week. This project was approved by the Institutional Review Board of Taipei City Hospitals (TCHIRB-10505112-E).
Outcome variables
The outcome variable of interest was treatment outcome, which was categorised into two groups: successful treatment and mortality. Mortality was classified as TB-specific or non-TB-specific death, which was determined by the Taiwan Death Certification Registry [9]. TB-specific death in this study was defined as the underlying cause of death being due to TB according to the Taiwan Death Certification Registry (International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes: A010–A018; ICD-10-CM code: A15–A19). Non-TB-specific death was defined as any underlying cause of death other than TB.
Explanatory variables
Covariates identified in previous studies as risk factors for TB mortality or as major comorbidities linked to TB mortality were assessed in the analyses [Reference Waitt10], including demographic factors (age, sex, marital status, education level, nursing home residence, smoking, alcohol use and employment status); clinical findings (chest radiography findings, acid-fast bacilli (AFB) smear status, drug resistance); underlying diseases (malignancy and ESRD); source of notification and mode of treatment. The source of the notification was defined as the department that reported the TB case, including ordinary wards, intensive care units or outpatient services. Treatment mode was categorised as directly observed therapy or self-administered therapy. Directly observed therapy was defined as anti-TB medication ingestion that was directly supervised by a trained observer [Reference Jasmer11]. Self-administered therapy referred to unsupervised treatment.
Statistical analysis
In univariate analysis, the χ 2 test was used to assess associations of the clinical relevant variables with the outcomes. All variables found to be significant (P < 0·05) through univariate analysis were considered for inclusion in multivariate analysis. Multiple logistic regression was used to assess the multivariate associations of the factors with all-cause mortality. Furthermore, multinomial logistic regression was used to identify the factors associated with TB-specific and non-TB-specific mortality. Adjusted odds ratios (AOR) with 95% confidence intervals (CI) were reported to show the strength and direction of these associations. All data management and analyses were performed using the SAS 9·4 software package (SAS Institute, Cary, NC).
RESULTS
Characteristics of patients with TB
During the 9-year study period, there were 5081 TB cases (age ⩾65 years) reported to the Taipei TB Control Department. Of these, 21 were lost to follow up, 16 were still on treatment, 1 had treatment failure, 11 had incomplete data and 21 had transferred out of Taipei City (Fig. 1). The remaining 5011 were included in the subsequent analysis. Overall, the mean and median ages of the study subjects were 79·7 years (range 65–113) and 80 years (interquartile range 74–85), respectively; 74·1% were men; 67·3% (3372) were successfully treated; 7·1% (354) died of TB-specific causes and 25·6% (1285) died of non-TB-specific causes during TB treatment.
Fig. 1. Study flow diagram. TB, tuberculosis.
The TB-specific mortality rates were 3·3%, 5·7% and 12·7% in patients aged 65–74, 75–84 and ⩾85 years, respectively. Additionally, the non-TB-specific mortality rates were 14·0%, 26·0% and 36·2% in patients aged 65–74, 75–84 and ⩾85 years, respectively.
Univariate analyses of factors associated with all-cause mortality
The χ 2 test revealed variables that were significantly associated with all-cause mortality, i.e. age 75–84 or ⩾85 years, unemployment, residence in a long-term care facility, ESRD, malignancy, AFB-smear positivity, TB-culture positivity, pleural effusion on chest radiograph and notification from an ordinary ward or intensive care unit (Table 1). Additionally, variables significantly associated with a lower chance of all-cause mortality were elementary school, high school, and university or higher education, a cavity on chest radiograph and directly observed therapy.
Table 1. Univariate analyses of risk factors for mortality in elderly TB patients, Taipei, Taiwan (2006–2014)
OR, odds ratio; CI, confidence interval; LTCF, long-term care facility; DM, diabetes mellitus; ESRD, end-stage renal disease; TB, tuberculosis; CXR, chest radiograph; MDR-TB, multidrug-resistant tuberculosis; SAT, self-administrated treatment; DOT, directly observed treatment.
*<0·05; **<0·01; ***<0·001.
Multivariates analyses of factors associated with all-cause mortality
All variables found to be significant (P < 0·05) through univariate analysis were considered for inclusion in multivariate analysis. After adjusting for socio-demographic factors and comorbidities, risk factors associated with all-cause mortality included age 75–84 years (AOR = 2·18, 95% CI 1·82–2·61), age ⩾85 years (AOR = 4·57, 95% CI 3·78–5·53), residence in a long-term care facility (AOR = 1·57, 95% CI 1·22–2·01), ESRD (AOR = 2·63, 95% CI 1·92–3·60), malignancy (AOR = 3·01, 95% CI 2·43–3·74), AFB-smear positivity (AOR = 1·59, 95% CI 1·37–1·85), TB-culture positivity (AOR = 1·53, 95% CI 1·29–1·81), pleural effusion on chest radiograph (AOR = 1·40, 95% CI 1·16–1·69) and notification from an ordinary ward (AOR = 2·50, 95% CI 2·17–2·88) or intensive care unit (AOR = 6·71, 95% CI 4·72–9·54) (Table 2). Additionally, protective factors for all-cause mortality included a high school education (AOR = 0·72, 95% CI 0·58–0·89), a university or higher education (AOR = 0·64, 95% CI 0·51–0·81), a cavity on chest radiograph (AOR = 0·76, 95% CI 0·61–0·95) and directly observed therapy (AOR = 0·34, 95% CI 0·29–0·39).
Table 2. Multivariates analyses of risk factors for all-cause mortality in elderly TB patients, Taipei, Taiwan (2006–2014)
TB, tuberculosis; AOR, adjusted odds ratio; CI, confidence interval; LTCF, long-term care facility; ESRD, end-stage renal disease; AFB, acid-fast bacilli; CXR, chest radiograph; DOT, directly observed treatment.
*<0·05; **<0·01; ***<0·001.
Factors associated with TB-specific and non-TB-specific mortality
Multinomial regression showed that, after controlling for other variables, risk factors associated with TB-specific and non-TB-specific mortality included age ⩾75 years, residence in a long-term care facility, ESRD, AFB-smear positivity, TB-culture positivity and notification from an ordinary ward or intensive care unit (Table 3). Protective factors for TB-specific and non-TB-specific mortality included directly observed therapy. Additionally, malignancy and pleural effusion on chest radiographs were significantly associated with a higher risk of non-TB-specific mortality; while high school, university or higher education and a cavity on chest radiograph were significantly associated with a lower risk of non-TB-specific mortality.
Table 3. Multinomial regression: demographic and clinical variables associated with TB-specific and non-TB-specific death among elderly TB patients in Taipei, Taiwan (2006–2014) a
TB, tuberculosis; AOR, adjusted odds ratio; CI, confidence interval; LTCF, long-term care facility; ESRD, end-stage renal disease; AFB, acid-fast bacilli; CXR, chest radiograph; DOT, directly observed treatment.
a Reference is successfully treated individuals.
**<0·01; ***<0·001.
DISCUSSION
In this large cohort study of 5011 TB cases, the overall proportion of death was 32·7% in 2006–2014. After controlling for potential confounders, age ⩾75 years, ESRD, malignancy, AFB-smear positivity, TB-culture positivity, pleural effusion on chest radiograph and notification by ordinary ward or intensive care units were associated with a higher risk of all-cause death; while high education, a cavity on chest radiograph and directly observed therapy were associated with a lower risk of all-cause death.
This study found that the mortality was extremely high in elderly patients with TB, which was higher than 3·9% in Taiwanese elderly population without TB [12]. As compared with patients aged 65–74 years, patients aged 75 years and older had an even higher risk of TB-specific and non-TB-specific mortality. Higher mortality among older TB patients may be due to waning immunity and increased comorbidities. Additionally, elderly patients with TB are more likely to experience a delay in TB diagnosis and treatment [Reference Storla13], which might cause the high mortality in this population. Since elderly individuals are at an increased risk of TB infection [1], clinicians need to be aware of TB in this population.
We identified individual socio-economic status variables that were risk factors for mortality. Specifically, individuals with at least a high school education were associated with a lower risk of death. Prior studies have reported that people with higher socio-economic status are more likely to receive superior treatment and additional diagnostic procedures [Reference Younis14]. Moreover, patients with low educational levels had a poor understanding of TB, which resulted in a delay in diagnosis [Reference Hoa15], interruption of treatment [Reference Elzinga16] and a higher rate of mortality [Reference Okanurak17]. To improve TB treatment outcomes in elderly individuals, future control programmes should particularly target patients with a lower educational level.
This study found that a cavity on chest radiograph was associated with a lower risk of mortality. The protective effect of cavitary diseases on TB mortality in this study may reflect the fact that cavitary diseases raise the suspicion for TB in clinicians’ minds and reduce diagnostic delay [Reference Lin18]. Since elderly patients with TB are more likely to have a non-specific clinical presentation at TB onset (e.g. no cavity lesions on chest radiograph) [Reference Cruz-Hervert4, Reference Chiang19], elderly patients suspected to have TB should be evaluated carefully to ensure early initiation of treatment.
This study showed that patients receiving directly observed therapy had a lower risk of mortality than those on self-administered therapy. In the Taipei DOTS programme, each directly observed treatment (DOT) observer monitors 5–10 patients with TB. DOT observers are trained to interview patients with TB about their TB symptoms and complications of treatment under the supervision of public health nurses. When patients with TB on DOT have, for example, worsened dyspnoea or blurred vision, public health nurses contact the doctors to arrange a hospital visit. Public health nurses daily check the regimen and dosage of TB drugs of each DOT patient according to TB treatment guidelines daily [Reference Hopewell20], and remind prescribing doctors, if needed, to ensure the patients are provided with appropriate TB drugs. DOT has been recommended for TB patients to improve treatment adherence [Reference Frieden21]. Our study suggests that DOT programme should be applied to all elderly patients with TB to further reduce mortality.
Consistent with previous reports, long-term care facility residences [Reference Yen23], malignancy [Reference Kliiman24], ESRD [Reference Kliiman24], AFB-smear positivity [Reference Pepper25], TB-culture positivity [Reference Kliiman24], pleural effusion on chest radiograph [Reference Kliiman24] and notification from an ordinary ward or intensive care unit [Reference Pepper25] were associated with a higher risk of mortality.
The strengths of this study included that this was a city-wide population-based study with a large number of patients enrolled in the cohort, and treatment outcomes were tracked and recorded using standardised systems. However, several limitations should be considered when interpreting the findings of this study. First, this was a secondary-data study. Important information about TB patients, e.g. intravenous drug use, was not available in the surveillance data. Second, the cause of death among TB patients relied on ICD-9-CM codes in the Taiwan Death Certification Registry, and the outcome of cause of death may have been misclassified. This non-differential misclassification of outcome would bias the results towards a null association. Finally, the external validity of our findings may be a concern because almost all our enrollees were Taiwanese. The generalisability of our results to other non-Asian ethnic groups requires further verification.
CONCLUSIONS
According to the above analysis, this study found that the proportion of death among elderly patients with TB in Taipei, Taiwan, was high. Age ⩾75 years, ESRD and malignancy were associated with a higher risk of all-cause mortality; while high education, a cavity on chest radiograph and directly observed therapy were associated with a lower risk of all-cause death. To improve TB treatment outcomes, future control programmes should particularly target individuals with comorbidities (e.g. ESRDs and malignancy) and those with a lower socio-economic status (e.g. not educated). Additionally, a directly observed therapy programme should be applied to all elderly patients with TB to further reduce mortality.
ACKNOWLEDGEMENTS
The authors are grateful to registered nurses Jen-Chieh Hsiao and Miao-Yun Chen for interviewing the subjects and reviewing their medical records.
DECLARATION OF INTEREST
None.
