The majority of studies in this area have involved the interviewing of patients via questionnaires and have collected patients' self-reported periods of time between the onset of TB symptoms and their first visit to a medical provider (patient delay) and between their first contact with a qualified doctor and the initiation of TB treatment (health system delay). This study is the first to analyze data obtained from a nationwide database, the TB reporting enquiry system. One limitation of this study is that data from which to estimate the time between the onset of coughing and the first medical examination are not available, and only limited data are available for analysis. The completeness of data reporting was not perfect in the early period of use of the reporting enquiry system, and missing data (unknown data) was therefore inevitable.
Adjusting for other variables, being female, in comparison with being male, had a protective factor with respect to treatment delay. This finding is similar to those of other studies conducted in Kampala  and southern India ; however, some other studies have identified being female as having adverse consequences in terms of a longer delay in diagnosis [6, 9] and in treatment [10–12].
Similar to the findings of many other studies, diagnosis and treatment delays were found to significantly increase with age [7, 9, 13–17]. In Taiwan, the life expectancy in 2006 for men and women was 75.1 and 81.9 years, respectively . In this study, the mean age was 63.3 years, and 54.6% of patients were aged over 65 years. Many studies [13, 18–20] have stratified and only studied those who were aged over 65, but because the sample size of this study was large enough, we chose to divide the rest of the patients into two groups, those aged under 35 and those aged between 35 and 65. A dose-response effect of age was found in this study, and the older the patient, the more serious the consequences of diagnosis and treatment delays.
Being of an aboriginal ethnic background was found to be a significant risk factor associated with treatment delay as compared with the general Taiwanese population (1.92 times higher). Taiwan has impelled the aboriginal community to attend health screening since 2001: aboriginal people aged 15 years and over are urged to take advantage of the free chest radiograph inspections offered every year, and diagnosis would then not be a problem. However, the medical service resources in aboriginal areas are relatively deficient, leading to delays in treatment time. It should be noted that the mortality rate and incidence of TB are much greater in aboriginal communities than in non-aboriginal areas in Taiwan [19, 21–24], and this may have an influence on prevention and treatment . In contrast, being from abroad was found to be a significant protective factor for diagnosis delay. Some studies have also shown that those from abroad have a shorter health care system delay [18, 26].
A total of 27,209 PTB patients had a positive sputum smear, and of these, 19.7% of sputum cultures were negative. One possible reason for this may be the density of the sputum collected. A positive sputum culture was found to be associated with increased delays only when the sputum smear was negative, not when it was positive. Some studies have found that a negative sputum smear is associated with greater diagnosis and treatment delays [14, 20, 27–29], and one study found that a negative sputum smear was negatively associated with diagnosis delay .
In this study, one possible reason for the opposite trends of diagnosis and treatment delay from 2002 to 2006 may be that the method of TB diagnosis was chest radiography prior to 2002, but gradually moved towards laboratory testing from 2002 onwards. The process of testing may postpone the time of diagnosis, but on the other hand, Taiwan CDC and the NHI cooperate with the policy that "the TB patient who is not notified does not pay," which prevents the hindering of treatment by doctors.
Living with family as compared with living alone resulted in a significantly shorter treatment delay but a 1.11 times higher diagnosis delay. Two previous studies found that being single was a positively-associated risk factor of diagnosis delay, namely studies performed in Ethiopia  and Botswana . Diagnosis at a non-medical center was found to result in a significantly shorter diagnosis delay but a 1.09 times higher treatment delay. Some previous studies have found that an initial visit to a hospital was a negatively-associated risk factor of diagnosis delay [7, 17, 31, 32]; furthermore, it has also been found that an initial visit to a governmental low-level healthcare facility, a traditional/unqualified practitioner or a private practitioner is a positively-associated risk factor of diagnosis delay [6, 16, 33, 34].