Health care seeking behavior and patient delay in tuberculosis diagnosis Comportamento de busca por cuidados em saúde e atraso do paciente no diagnóstico da tuberculose Comportamiento de búsqueda de atención de salud y retraso del paciente en el diagnóstico de la tuberculosis

Delays in diagnosis of TB cases are major impeding factors in the control of TB. The objectives of this study were to describe the health care seeking behavior of TB patients, assessing patient delay and the number of health care facilities visited before the start of TB treatment. A crosssectional study was carried out with adult patients with pulmonary TB presenting to two TB facilities to start treatment. We found a median patient delay of 20 days. The factors associated negatively with patient delay in multivariate analysis were weight loss, and have sought treatment because of the first symptom. We also demonstrated that 44.8% of patients incorrectly reported the mode of transmission of TB. In addition, the local of first attendance was an emergency room of public hospitals in 37.3% of patients. We demonstrated that the median patient delay in TB diagnosis in two TB services in a region with a high prevalence of TB was 20 days, and the protective factors associated with this delay in multivariate analysis were weight loss, and have sought treatment because of the first symptom. Tuberculosis; Delayed Diagnosis; Time-to-Treatment Resumo Atrasos no diagnóstico de casos de tuberculose são os principais fatores que impedem o controle da doença. Os objetivos deste estudo foram descrever a trajetória de pacientes com tuberculose, avaliando a demora do paciente e o número de unidades de saúde visitadas antes do início do tratamento. Estudo transversal foi conduzido com pacientes com idade ≥ 18 anos e com diagnóstico de tuberculose pulmonar, que responderam a um questionário. A média de idade dos pacientes foi de 40,4 ± 16,1. Foi encontrada uma mediana de 20 dias no atraso dos pacientes. Também foi verificado que 44,8% relataram incorretamente o modo de transmissão da tuberculose. Além disso, o local do primeiro atendimento para 37,3% dos pacientes foi uma emergência de hospital público. Encontrou-se uma mediana de 20 dias no atraso dos pacientes até o diagnóstico da tuberculose; os fatores protetores associados a este atraso na análise multivariada foram perda de peso e ter procurado tratamento por causa do primeiro sintoma. Tuberculose; Diagnóstico Tardio; Tempo para o Tratamento 321 ARTIGO ARTICLE http://dx.doi.org/10.1590/0102-311X00195413 Almeida CPB et al. 322 Cad. Saúde Pública, Rio de Janeiro, 31(2):321-330, fev, 2015 Introduction Tuberculosis (TB) is a major public health issue worldwide, particularly in lowand middle-income countries 1. Delays in diagnosis and treatment of TB cases are major impeding factors in the control of TB. A single infectious person who remains untreated can infect between ten and fifteen people every year, spreading the infection in the community 1,2. In addition, delayed TB diagnosis and treatment may result in more extensive disease and more complications, and increases the risk of mortality 3,4,5. The total delay in TB diagnosis can be divided into the length of patient delay and the length of health care system delay 6. Socio-economic status, perceived severity of illness and symptom recognition, distance and physical access, and perceived quality and cost of health service provision are all factors associated with patient delay 7. Previous studies have shown that a delay in diagnosis is closely related to poor access to health care 8,9. Despite TB control programs’ recommendation that diagnosis should be made at the primary health care level, most patients are still diagnosed in hospitals. In fact, in Porto Alegre, Rio Grande do Sul State, Southern Brazil, 38.98% of TB cases were reported by hospitals 10. Understanding the factors related to delay in diagnosis of disease is essential to reduce the period of transmission and the risk of exposure of other community members. The objectives of this study were to describe the health care seeking behavior of TB patients, assessing patient delay and the number of health care facilities visited before the start of TB treatment. Materials and methods We conducted a cross-sectional study in two TB services, located in the city of Porto Alegre, between February and December 2012. Porto Alegre is the city with the highest incidence of TB in Brazil (115 cases/100,000 population), with a cure rate of 67%. Also, the TB-HIV co-infection rate is the highest in the country (35%) 11. All adult patients (≥ 18 years old) with pulmonary TB presenting to the two facilities that admit TB patients to start treatment were invited to participate. Pulmonary TB was diagnosed according to any of the following criteria established in the Brazilian Guidelines for Tuberculosis 12. Patients who refused signing the consent form were excluded from the study. Face-to-face interviews were performed using a standardized and pre-tested questionnaire developed exclusively for this study to collect information on each patient at the time of admission. The interviews were conducted by the same interviewer, previously trained. Medical records were also reviewed. The following data were recorded: demographic data, symptoms before seeking health care, health care seeking behavior, type and number of health care facilities visited, and knowledge about TB. Health care seeking time was defined as the duration (in days) between the onset of symptoms to the first visit to a health care facility (including hospital, TB services, primary health care clinics, etc). In order to evaluate the associated factors with patient delay, the study population was divided into two groups, according to the median time of health care seeking: group 1 (≥ median) and group 2 (< median). Individuals included in group 1 were considered to have “patient delay”. Data analysis was performed using SPSS 18.0 (IBM Corp., Armonk, United States). Data were presented as number of cases, mean ± standard deviation (SD), or median with interquartile range (IQR). Categorical comparisons were performed by chi-square test using Yates’s correction if indicated or by Fisher’s exact test. Continuous variables were compared using the t-test or Wilcoxon test. Multivariate logistic regression analysis was used to evaluate risk factors for patient delay, using selection of factors associated (p ≤ 0.20) with delay in univariate analysis or those known to have clinical significance. Stepwise regression models were also examined to evaluate the possible collinearity among the predictors. The predictors selected in the final model were based on both statistical and clinical significance. The goodness of fit of the multiple logistic regression models was assessed using the Hosmer-Lemeshow test. Odds ratios (ORs) and nominal 95% confidence intervals (95%CIs) were presented. A two-sided p-value < 0.05 was considered significant for all analyses. On the basis of data from van der Werf et al. 13, who informed that individuals reporting unemployment, cough or loss of weight before seeking health care had a longer patient delay, we estimated a sample size of 134 patients, under the assumptions of a type I error (two-sided) of 5% and a type II error of 10%.


Abstract
Delays in diagnosis of TB cases are major impeding factors in the control of TB.The objectives of this study were to describe the health care seeking behavior of TB patients, assessing patient delay and the number of health care facilities visited before the start of TB treatment.A crosssectional study was carried out with adult patients with pulmonary TB presenting to two TB facilities to start treatment.We found a median patient delay of 20 days.The factors associated negatively with patient delay in multivariate analysis were weight loss, and have sought treatment because of the first symptom.We also demonstrated that 44.8% of patients incorrectly reported the mode of transmission of TB.In addition, the local of first attendance was an emergency room of public hospitals in 37.3% of patients.We demonstrated that the median patient delay in TB diagnosis in two TB services in a region with a high prevalence of TB was 20 days, and the protective factors associated with this delay in multivariate analysis were weight loss, and have sought treatment because of the first symptom.

Introduction
Tuberculosis (TB) is a major public health issue worldwide, particularly in low-and middle-income countries 1 .Delays in diagnosis and treatment of TB cases are major impeding factors in the control of TB.A single infectious person who remains untreated can infect between ten and fifteen people every year, spreading the infection in the community 1,2 .In addition, delayed TB diagnosis and treatment may result in more extensive disease and more complications, and increases the risk of mortality 3,4,5 .
The total delay in TB diagnosis can be divided into the length of patient delay and the length of health care system delay 6 .Socio-economic status, perceived severity of illness and symptom recognition, distance and physical access, and perceived quality and cost of health service provision are all factors associated with patient delay 7 .
Previous studies have shown that a delay in diagnosis is closely related to poor access to health care 8,9 .Despite TB control programs' recommendation that diagnosis should be made at the primary health care level, most patients are still diagnosed in hospitals.In fact, in Porto Alegre, Rio Grande do Sul State, Southern Brazil, 38.98% of TB cases were reported by hospitals 10 .Understanding the factors related to delay in diagnosis of disease is essential to reduce the period of transmission and the risk of exposure of other community members.
The objectives of this study were to describe the health care seeking behavior of TB patients, assessing patient delay and the number of health care facilities visited before the start of TB treatment.

Materials and methods
We conducted a cross-sectional study in two TB services, located in the city of Porto Alegre, between February and December 2012.Porto Alegre is the city with the highest incidence of TB in Brazil (115 cases/100,000 population), with a cure rate of 67%.Also, the TB-HIV co-infection rate is the highest in the country (35%) 11 .
All adult patients (≥ 18 years old) with pulmonary TB presenting to the two facilities that admit TB patients to start treatment were invited to participate.Pulmonary TB was diagnosed according to any of the following criteria established in the Brazilian Guidelines for Tuberculosis 12 .Patients who refused signing the consent form were excluded from the study.
Face-to-face interviews were performed using a standardized and pre-tested questionnaire developed exclusively for this study to collect information on each patient at the time of admission.The interviews were conducted by the same interviewer, previously trained.Medical records were also reviewed.The following data were recorded: demographic data, symptoms before seeking health care, health care seeking behavior, type and number of health care facilities visited, and knowledge about TB.
Health care seeking time was defined as the duration (in days) between the onset of symptoms to the first visit to a health care facility (including hospital, TB services, primary health care clinics, etc).In order to evaluate the associated factors with patient delay, the study population was divided into two groups, according to the median time of health care seeking: group 1 (≥ median) and group 2 (< median).Individuals included in group 1 were considered to have "patient delay".
Data analysis was performed using SPSS 18.0 (IBM Corp., Armonk, United States).Data were presented as number of cases, mean ± standard deviation (SD), or median with interquartile range (IQR).Categorical comparisons were performed by chi-square test using Yates's correction if indicated or by Fisher's exact test.Continuous variables were compared using the t-test or Wilcoxon test.Multivariate logistic regression analysis was used to evaluate risk factors for patient delay, using selection of factors associated (p ≤ 0.20) with delay in univariate analysis or those known to have clinical significance.Stepwise regression models were also examined to evaluate the possible collinearity among the predictors.The predictors selected in the final model were based on both statistical and clinical significance.The goodness of fit of the multiple logistic regression models was assessed using the Hosmer-Lemeshow test.Odds ratios (ORs) and nominal 95% confidence intervals (95%CIs) were presented.A two-sided p-value < 0.05 was considered significant for all analyses.
On the basis of data from van der Werf et al. 13 , who informed that individuals reporting unemployment, cough or loss of weight before seeking health care had a longer patient delay, we estimated a sample size of 134 patients, under the assumptions of a type I error (two-sided) of 5% and a type II error of 10%.

Results
One hundred thirty-nine patients met the inclusion criteria.Five patients refused to participate, then one hundred thirty-four patients were included in the analysis.The mean age of all pa-tients was 40.4 ± 16.1 years, 62.7% were males, and 59% were white.The majority of patients were literate and unemployed.Twenty-three patients (17.1%) were HIV positive.The characteristics of participants are summarized in Table 1.
Health seeking behavior, knowledge and perspectives of TB patients were described in Table 2.One hundred and seven patients (79.9%) sought treatment after the first symptom of TB.Among those who did not seek treatment soon after the onset of the first symptom, 44.8% did not because they thought the symptom was normal, 21.6% thought would improve without treatment, and 7.5% said that they could not miss work to sought treatment.One hundred and two patients (76.1%) sought care elsewhere before TB health center.The type of facility first sought by patients was: emergency of a public health hospital (n = 50, 49.1%), public health center (n = 35, 34.3%), private practice (n = 11, 10.8%), and emergency of a private hospital (n = 6, 5.9%).Overall, the median number of visits to health care facilities before the diagnosis was 3 (IQR: 2-4).Thirty seven patients (27.6%) were diagnosed and started treatment at hospitals, and did not seek treatment at health centers.The reasons for not seeking treatment in health centers were: "it is difficult to get medical care there" (n = 15), "I thought my problem would not be solved there" (n = 12), "I don't know where is the health center I should go" (n = 7), and "I did not know I could seek a health center for this problem" (n = 3).
The median health care seeking time was 20 days (IQR: 7-46.3 days).According to this median time, 71 (52.9%) patients were included in group 1 (≥ 20 days), and 63 (47%) in group 2 (< 20 days).Table 3 summarizes the results of the univariate analysis.By univariate analysis, characteristics associated negatively with delayed initial health seeking were weight loss, sought care after the first symptom, and smear positive sputum.
Logistic regression analysis estimating the ORs of risk of patient delay was conducted.Age, sex, monthly household income < 1 stan-

Discussion
In this study, we found a median health care seeking time of 20 days.The protective factors for patient delay in multivariate analysis were weight loss, and have sought treatment because of the first symptom.We also demonstrated that approximately 50% of patients incorrectly reported the mode of transmission of TB.In addition, the local of first attendance was an emergency room of public hospitals in almost 40% of patients.
Identifying the sources of delay is a critical issue for an effective TB control.Patient delay may explain why many patients present with advanced disease and why a relatively high percentage of TB patients die.Furthermore, patient delays lead to increased spread of TB within the community 14,15,16 .There is no agreed definition of what constitutes an "acceptable" delay, and it probably depends on the local health services and the local epidemiological situation, with a shorter delay to be expected when incidence is high 17 .Care seeking or patient delay is widely variable in literature, ranging from 4.9 to 162 days 18 .In our study, patient delay was similar to that found by other authors in India 19,20 , Japan 21 , Hong Kong 22 , and Spain 23 .
Persons with suspected TB symptoms but who do not seek health care should be identified, and reasons for not seeking treatment need to be investigated.Having sought help after the appearance of the first symptom was a protective factor for patient delay in our study.The main reasons for delay among those who presented late to the health system involve financial, psychosocial and cultural characteristics of patients 24,25 .
We identified among those who did not seek treatment soon after the onset of the first symptom, that the main justifications were because they thought the symptom was normal, they thought would improve without treatment, and that they could not miss work to sought treatment.Similar reasons were described in another study: patients "thought symptoms would go away" and "symptoms not considered serious" 13 .It has also been reported that patients awaiting symptoms resolution without specific treatment and can assign them to viral infections 26,27,28,29,30 .
Our results showed that weight loss was a protective factor for patient delay.In a recent study 31 , authors demonstrated that the presence of non-specific symptom of cough was a risk factor associated with longer delays, but if other pulmonary symptoms like loss of weight, for example, were considered together these delays could be shorten.Another previous study revealed that persistent cough not accompanied by symptoms such as weight loss and weakness could delay patients' first initiative to seek help 32 .It is possible that patients' knowledge that weight loss is one of the TB symptoms raised awareness of TB among them.However, the patients in our study demonstrate a lack of knowledge about TB, once almost one half did not know how the disease was transmitted.Misconceptions about TB transmission have been reported in other studies 33 , and it is related to delays in diagnosis and treatment non-adherence 33,34,35,36 .We could not show an association between patient delay and lack of knowledge about TB, but is important to develop strategies to improve patient education about their illness in the TB services, since it can have implications in treatment adherence.In the same way, it is important to train health professionals in identifying these patients.One study 37 conducted in Vitória, Espírito Santo State, Brazil, found that the majority of respiratory symptomatic patients (71%) had not sought health care facility because of cough, the symptom most commonly associated with TB.
The median number of visits to health care facilities before the diagnosis was 3, and the number of sputum smears requested in those facilities was very low (less than 40%).Loureiro et al. 38 also demonstrated that sputum smears were requested in 42.1% of patients who sought care in primary care level.This finding could be related to lack of awareness of TB among health care workers 39 .Another concerning finding of the present study was that the local of first attendance was an emergency room of public hospitals in almost 40% of patients.In addition, almost 30% were diagnosed and started treatment at hospitals, and did not seek treatment at health centers.The reasons for not seeking treatment in health centers were mainly poor access to medical care in such places, and lack of credibility in the treatment that can be offered there.In addition, lack of resoluteness of patient's symptoms in the primary care level is an issue to be considered.Another fact to be considered is the greater flexibility of office hours in hospitals as compared with health centers 40 .One study recently reported that most patients treated for TB in a primary health clinic in a high TB incidence city in Southeast Brazil, initially sought medical attention at an emergency room rather than the primary health care clinic, as recommended by the public health system in Brazil 41 .In another study conducted in Vitoria, Brazil, 30.7% of patients initially seek treatment at an emergency room 38 .
Emergency room is one of the main entrances for public health care system in Brazil, where a significant number of individuals with respiratory symptoms like cough go in order to get the first aid.Considering that emergency room is a site with a high potential for propagation of TB, efforts should be made for early diagnosis and treatment.Furthermore, the reasons for patients seeking emergencies instead of health centers should be better investigated.Some considerations should be taken into account when interpreting the results of this study.First, patient delays were determined retrospectively then they could have been limited by recall bias.Nevertheless, our findings indicate that there is a significant time elapsed before patients sought medical care.Second, we evaluated only patient delay, and not health care system delay, that could have a role in delay.Although delayed initial health seeking covers only a part of the total delay, it can impact on transmission.A study conducted in Foz do Iguaçu, Paraná State, Brazil showed that the time between first consultation and diagnosis was not the main problem in total time delayed, and patient delay was more important 42 .Additionally, the difficulty of establishing the boundaries between the patient delay and the health system delay was previously discussed, since previous experiences may influence patient's decision to seek the service again or delay the search for care 42 .Also, we chose a cutoff of 20 days (median health care seeking time) that is near the time reported in the media for seeking care (cough ≥ 2 weeks); if we had chosen another cutoff, we might have found other associations.However, a similar cutoff was already used in a study in Brazil 43 .In addition, the use of a previously validated questionnaire 44 would facilitate comparison with data from other studies.Nevertheless, the use of a specific questionnaire was useful for characterizing these patients, identifying regional differences in health assistance and factors associated with delayed diagnosis in this population.Eventually this knowledge will be helpful to adequate planning of future actions in the area studied.Finally, we did not evaluate the impact that patient delay may have on mortality.In spite of these concerns, we recruited patients from two TB services, and we expect the results may thus be applied to other similar settings.
In conclusion, we demonstrated that the median health care seeking time in two TB services in a region with a high prevalence of TB was 20 days, and the factors associated with patient delay in multivariate analysis were weight loss, and have sought treatment because of the first symptom.Also, patients studied showed a lack of knowledge of TB, and a high percentage of them first sought attendance in emergency rooms of public hospitals.Future studies on reasons for not seeking care in health centers or TB services may be warranted in this setting.In addition, our results showed that there is a need for public health interventions, focusing on TB education programs, including awareness of early symptoms of disease.

Resumen
Los retrasos en el diagnóstico de la tuberculosis son los principales factores que impiden el control de la tuberculosis.Entre los objetivos de este estudio se encuentra la descripción de la trayectoria de los pacientes con tuberculosis, evaluando el retraso en el diagnóstico del paciente y el número de establecimientos de salud visitados antes de iniciar el tratamiento para la tuberculosis.Se realizó un estudio transversal con pacientes ≥ 18 años de edad y con diagnóstico de tuberculosis pulmonar que respondieron a un cuestionario.La edad media de los pacientes fue de 40,4 ± 16,1.Se encontró un media de 20 días en el retraso del diagnóstico a los pacientes.También se halló que un 44,8 % informó incorrectamente del modo de transmisión de la tuberculosis.Por otra parte, el primer punto de asistencia para un 37,3 % de los pacientes fue el servicio de urgencias en un hospital público.Se halló un promedio de 20 días en el retraso del diagnóstico de la tuberculosis a los pacientes, los factores protectores asociados con el retraso en el análisis multivariado fueron la pérdida de peso y conseguir el tratamiento debido cuando se presenta el primer síntoma.Tuberculosis; Diagnóstico Tardío; Tiempo de Tratamiento Contributors C. P. B. Almeida designed the study, collected data, and help to draft the manuscript.E. K. Skupien collected and analyzed data, and help to draft the manuscript.D. R. Silva designed the study, analyzed data, and wrote the paper.

Table 2
Health seeking behavior, knowledge and perspectives of TB patients (n = 134).
dard minimum wage, weight loss, night sweats, fever, sought treatment after the first symptom, and previous TB were included in the multivari-ate analysis.The variables associated with patient delay in multivariate logistic regression are shown in Table4.

Table 3
Univariate analysis of factors associated with patient delay.

Table 4
Multivariate analysis of factors associated with patient delay.