This is the first study from Mozambique that assessed the prevalence of and the risk factors associated with patient and health system delays. The median total delay of 150 days was similar to the delay observed in Tanzania (136 days)  and in Ghana (120 days) . However, it was much higher than the delay observed in other African countries such as Uganda (84 days) , South Africa (60 days) , rural Botswana (84 days) , and Nigeria (70 days) . This finding is particularly concerning as the majority of the patient delay and health system delay studies carried out elsewhere were conducted in referral hospitals, where longer patient and health system delays are expected. Our study was carried out in primary health clinics. The fact that patient delay and health system delay had a similar contribution to total delay indicates that there are many opportunities to reduce total delay by intervening both at the community and in the health care system.
The median patient delay of 60 days observed in our study was similar to that observed in Addis Ababa, Ethiopia  and Lagos, Nigeria . However, it was higher than the delay observed in Uganda , South Africa , and Botswana .
The association between patient delay and farming observed in this study is consistent with findings from studies carried out elsewhere [11, 29]. It has been suggested that the increased delay observed among farmers may be related to their socio-economic condition, specifically low levels of education and high levels of poverty . The delay observed among farmers in our study may be related to the time farmers spend away from the city working in their fields. In addition, farming is a non-salaried and seasonally dependant work and this may force farmers to delay treatment seeking in the busy times even if they are very ill.
The increased delay among patients visiting first a traditional healer is also consistent with findings from other studies [19, 26, 32–34]. Our results indicate that traditional healers play an important role in the provision of TB care in this population, as a quarter of patients referred to traditional healers as the first contact of care and a significant proportion of patients indicated that TB is a traditional disease. This also indicates that there is a high level of misconceptions about TB in this population.
The increase in patient delay associated with poor TB knowledge has also been described in other studies [5, 18, 27, 31]. In this study, 37.6% of the patients said that they delayed seeking health care because they thought their symptoms would disappear. This finding is consistent with findings from other studies [12, 22, 31] and shows that some patients wait until symptoms become severe before seeking health care.
The increased delay among patients with an underlying chronic disease can be explained by the failure to recognize TB symptoms. Patients may relate the symptoms to their underlying chronic condition. The insidious character of TB may also contribute to the delay in this group of patients.
HIV infection was associated with shorter patient delay, a finding similar to a study carried out in Thailand . It is possible that patients with HIV infection are more educated about TB, as TB information is actively provided to all patients attending HIV-related services. In fact, in this study the proportion of patients with adequate TB knowledge was higher among HIV-positive patients compared to HIV-negative patients.
Health system delay
The median health system delay of 62 days was long and similar to that reported in Ghana  and Uganda . The duration of health system delay was long even among new smear-positive patients (with a median of 52 days), indicating a failure of the health care system to promptly initiate TB treatment.
The increased health system delay observed among farmers in our study may be explained by the difficulties in visiting the health facilities for follow up visits and to collect the results of diagnostic tests. The association between multiple visits to a health facility and health system delay have been described in several studies [16, 26, 29, 34, 36]. Low level of clinical suspicion , poor satisfaction with the health care received , and delays in obtaining proper laboratory analysis  are some of the possible explanations for the delay associated with repeated visits to health facilities. In addition, patients with multiple visits tend to visit different health providers in different health facilities [26, 29] making it difficult to reach a prompt TB diagnosis. It is difficult to ascribe cause and effect between delay and multiple visits, as multiple visits necessarily requires the passage of time between the visits. Despite that, multiple visits to a health facility provides a good indication that the healthcare system is failing to diagnosing TB as health workers should be able to suspect TB in any patient with chronic cough and a history of multiple visits to health facilities.
Increased health system delay was also associated with coexistence of an underlying chronic condition. Patients with multiple clinical conditions may present with complex symptoms making it difficult for the health worker to reach a TB diagnosis. This highlights the importance of suspecting TB in any patient with chronic respiratory symptoms, even if the patient has a known underlying condition. It also highlights the need for better TB diagnostic tools.
The present study had some limitations that should be taken in consideration in the interpretation of results. Most importantly, the duration of patient delay may have been underestimated due to poor patient recall of the onset of symptoms, their duration and the date of the first health-seeking encounter. In addition, TB is a chronic disease with an insidious start, making it difficult for patients to remember exactly when the symptoms started.