Because of increasing life expectancy and declining birth rates, the ageing population problem has become a critical worldwide public health concern, particularly in developed countries [1]. Dysfunction in cellular immunity caused by chronic comorbidities, malnutrition, and age-related changes can render elderly people more susceptible to infectious agents, such as Mycobacterium tuberculosis [2, 3]. In 2015, 10.4 million people were diagnosed as active tuberculosis (TB) and among them, 1.8 million died [4]. In industrialised societies, the trend of institutionalised care further exposes elderly patients to a higher risk of TB infection. The elderly population therefore represents a large reservoir of TB infection. In developing countries, TB continues to affect all susceptible individuals, including elderly adults [5].

Delay in initiation of anti-TB treatment is a major impediment to effective control of TB [6]. However, in elderly people, the clinical presentations of TB can be myriad and easily confused with other age-related illnesses [7]. Although the standard four-combined anti-TB treatment is highly effective, it is associated with a high pill burden, long treatment course, and severe drug-related adverse events. Elderly patients are more prone to experience adverse events such as severe hepatotoxicity during anti-TB treatment [8]. These factors result in a unique challenge and suboptimal outcomes in management of TB among the geriatric population.

Sputum smear microscopy remains the most common method for diagnosing pulmonary TB (PTB), but smear-positive TB accounted for only 56% of all notified new TB cases [4]. A mycobacterial culture, although more sensitive, requires an average of 9.7 and 20.2 days to detect M. tuberculosis in liquid and solid culture media, respectively [9]. The advent of rapid molecular diagnostic tools, which are sensitive, specific, and quick, provides new opportunities to facilitate the microbiological diagnosis of PTB [10, 11].

In this nationwide retrospective cohort study, we investigated the impact of advanced age (≥80) on delay and outcome of anti-TB treatment with an emphasis on the influence of rapid molecular diagnostic tools.

The National Health Insurance (NHI) programme of Taiwan is a compulsory insurance system covering 99.6% of the national population with a benefit package including comprehensive inpatient and outpatient medical services. The claims data were collected systemically and de-identified before being released for research purposes. The data were issued by the National Health Research Institute with delegation of authority from the Ministry of Health and Welfare under license for the current study.

In this study, patients with PTB during 2004–2009 were selected from the NHI Research Database (NHIRD) and followed-up until death or 31st December 2010, whichever came first. The Institutional Review Board of National Taiwan University Hospital approved the study (NTUH REC: 201,309,064 W).

From the nationwide database, 81,081 adult patients with PTB were identified (Fig. 1). Among them, 3747 (4.6%) died before anti-TB treatment began. The median age was 65.2 (47.5–76.9) years, with a male–female ratio of 2.25. The incidence rate of PTB was exponentially correlated to the age (R ² = 0.962; Fig. 2).

The elder patients required more outpatient visits, emergency room visits, admissions, chest x ray, mycobacterial culture, and MTB-NAAT studies to confirm the TB diagnosis and to start the anti-TB treatment (Additional file 1: Table S3). Among the ageing population, the duration of anti-TB treatment as well as duration covered by each first-line anti-TB drug was shorter while second-line anti-TB drugs were prescribed more frequently; indicating regimen modification. The elder patients had higher baseline disease severity reflected by the higher probabilities of requiring hospitalisation, intensive care, and mechanical ventilatory support during the first 14 days of anti-TB treatment, and consumed more medical resources during the anti-TB treatment course. However, the elder patients carried a lower treatment completion rate and a higher mortality rate (Fig. 3, Table 1). Female patients were younger, had less comorbidities, and a lower mortality rates than male patients (Additional file 1: Table S1).

Additional file 1: Table S2 summarises specific events that indicated the PTB onset. The prescription of airway medications or antibiotics tended to occur much earlier than chest x ray examination as well as mycobacterial study in the clinical course. The treatment delay was longer among the elderly than that among younger patients (Fig. 2). A later TB diagnosis year was associated with a shorter treatment delay in all age groups (P < 0.001) except for patients ≥80 years old (P = 0.678). The treatment delay was longer among patients with a fatal outcome (71 [11–139] days, Fig. 4) than those who completed treatment (33 [4–99] days) (P < 0.001)

Subpopulation analyses illustrated a stronger impact of performing MTB–NAAT on the treatment delay among the elderly patients (coefficient − 3.97 [−7.09, −0.84]) and smear-negative PTB (coefficient − 4.35 [−6.14, −2.55]) (Additional file 1: Table S4). In the sensitivity analysis adopting a stricter definition of treatment delay, the results were consistent with those in the main scenario (Additional file 1: Tables S5-S6, Table 3).

The present study is the first nationwide report on the outcome of anti-TB treatment in advanced age. It has three crucial findings. First, the incidence rate of PTB is exponentially correlated with the age that elderly adults are the major reservoir of PTB infections in Taiwan. Second, although the anti-TB treatment completion rate has increased following the implementation of directly observed treatment, short course (DOTS) programme in Taiwan since 2006, elder patients with PTB remained to have longer treatment delays and worse outcomes, particularly those with underlying comorbidities. Third, the length of treatment delay is inversely correlated with the treatment completion rate. The treatment delay can be shortened by applying rapid molecular diagnostic tools such as the MTB–NAAT. The extent of benefit is even greater among the elder patients and those with smear-negative PTB. Given the increasing elderly populations worldwide, the findings of the present study can serve as a reference for policies regarding TB care.

According to a World Health Organization report, the TB notification rate increases with age worldwide [4]. As a result of population ageing, the proportion of elder TB patients increased steadily from 1990 to 2015 [17]. The gradual deterioration of the immune system (involving both the host’s capacity to respond to infections and the development of long-term immune memory as age increases, referred to as immunosenescence) may be the major contributor [18, 19]. Other factors, such as malnutrition, poverty, decreased access to health services, comorbidities, and iatrogenic immunosuppression, also contribute to the higher risk of infection in ageing populations [20–22]. However, the correlation between age and PTB incidence has never been calculated, and reports on advanced aged population are currently lacking. This is the first study showing that the risk of PTB not only increases but is exponentially correlated with age (R ² = 0.962; Fig. 2).

Because of the high prevalence of underlying comorbidities, anti-TB treatment in elderly patients is frequently complicated by drug–drug interaction and adverse drug reactions, leading to an increased rates of regimen modification and default [8, 23]. Consequently, advanced age increases the mortality rate of TB significantly and eclipses the treatment completion rate [23–27]. In this study, the treatment completion rate among patients ≥65 years old was comparable to the two previous reports (71%–73%) [25, 26]. An even lower treatment completion rate was demonstrated among those with age ≥ 80 years.

Another crucial contributor to poor outcomes in elderly patients with PTB is the delay in anti-TB treatment. The clinical symptoms and radiographic findings of PTB in elderly people tend to be less specific [25, 28, 29]. Extrapulmonary TB including TB meningitis, osteomyelitis or urological involvement is more common with advancing age [3]. Combined with decreased access to health services [30], the atypical manifestations of TB in elder people result in a delay in the diagnosis and treatment [24, 28, 29]. As shown in the present study, prescription of airway medications and antibiotics occurred early in the course prior to chest x ray examination as well as the diagnosis of PTB, suggesting that these cases are already symptomatic and may be infectious in the community and health care system for a long period. Moreover, even when chest radiography is ordered, the duration of treatment delay is still far from negligible, indicating that a high proportion of patients presented with non-diagnostic radiographic findings, particularly in elderly patients. Consistent with previous studies, treatment delay increases mortality rates in patients with PTB [31, 32].

A treatment delay may result from either a delay in seeking health service (patient delay) or failure in establishing diagnosis and starting treatment (provider delay) [6, 33]. In countries with a high TB burden, insufficient patients’ awareness for the TB disease and financial barrier are major contributors for delay in diagnosis [6]. In countries with a low TB burden, the percentage of advanced pulmonary TB with positive sputum smear and cavitary lesions steadily increased due to declining clinicians’ vigilance to the presentations of TB and a lack of efficient diagnostic tools to diagnose TB in its early stage [34, 35]. Because of the built-in shortage of claims data, patient delay cannot be accessed in this study. However, the median of provider delay among patients aged 65–79 years was 32 days longer than that among those aged <65 years (53 vs 21 days). For those aged ≥80 years, the impact can be higher since the treatment delay is longer. In addition to the negative impact on treatment outcome, failure to recognise active PTB cases increases the risk of transmission [36], thus constituting a major hindrance to effective control for TB.

Because ageing is a well-known risk factor for adverse events during anti-TB treatment [8, 23, 37], for safety concerns, physicians are becoming increasingly hesitant to initiate anti-TB treatment unless solid bacteriologic evidence exists. Furthermore, because of the improved accessibility of health services in Taiwan, patients tend to seek medical help while their disease is minimal. This probably explains why an initial medical visit in an urban area is associated with a longer treatment delay than in a rural area. Implementing MTB–NAAT was shown to reduce treatment delay (Table 3), especially among the elderly and smear-negative PTB cases. These findings support the implementation of a rapid molecular assay for PTB diagnosis.

Most patient factors leading to treatment non-adherence can be eliminated with supervision, resulting in an improved treatment completion rate and reduction in unfavourable outcomes [38]. Under Taiwan’s national TB programme, DOTS has been implemented countrywide since 2006. The findings of this study support the continuous government commitment to TB control and the necessity of continuing DOTS programme in Taiwan.

The present study has some limitations. First, because of the built-in shortage of claims data, the results of mycobacterial studies and radiographic findings were unavailable. Second, the disease severity, a critical determinant of patient outcome, was not known. Although hospitalisation and admission to intensive care unit were used as surrogates of disease severity in this study, they may not correlate 100%. Third, the impact of the MTB–NAAT on treatment delay may be confounded by the indication, resulting in an overestimation of it benefits. Lastly and importantly, though the overall delay was calculated by fulfilling two or more events indicating TB onset, they could be due to clinical conditions other than TB. However, it may not be a considerable bias since sensitivity tests showed that the model was consistent across the broader or stricter definitions of delay in treatment.

The incidence of PTB increased exponentially with age. Ageing is associated with unfavourable outcomes and longer treatment delay, particularly for those with underlying comorbidities. Rapid molecular diagnostic tools can shorten treatment delay and should be integrated in the diagnostic algorithm for PTB, particularly in patients with advanced age.