Factors associated with tuberculosis cases in Semarang District, Indonesia: case–control study performed in the area where case detection rate was extremely low

Indonesia is ranked as the 4th highest contributor to tuberculosis (TB) in the world. Semarang District in Central Java displays extremely low case detection rate (CDR), possibly contributing to the local prevalence of TB. A case–control study was performed to explore the factors that cause such low CDR. We recruited 129 TB cases and 83 controls that visited the same centers and were not diagnosed with TB. The cases had 7.5 ± 2.3 symptoms/person on average, indicating the delay in diagnosis because the controls only displayed 1.0 ± 1.7. The multiple logistic regression analysis comparing the cases/controls extracted following factors as a risk to have TB: farmer, close contact with TB patients, ignorance of whether Bacillus Calmette-Guérin (BCG) was accepted or no, smoking, low income, a lot of people living in the same room, irregular hand wash before meals, not wash hands after blow, soil floor, and no sunlight and no ventilation in the house. Neither the cases nor the controls knew the symptoms and how to avoid TB infection, which probably caused the delay in diagnosis. It is difficult to change the current living conditions. Thus, the amendment of the community-based education program of TB seems to be required.

The 129 cases included 65 male and 64 female, respectively, and the average age was 41.2 ± 117 15.3 (Table 1). Although all the cases were registered at the health centers, around 20 % were 118 diagnosed at the different medical facilities (Table 2). Sputum smear was the first choice for 119 diagnosis [1,14], but 20 cases were diagnosed without any clinical examination. Chest X-ray 120 was utilized for the diagnosis of more than half of the cases. Long lasting cough with sputum 121 was the most common symptom that was followed by chest pain, malaise, anorexia, and 122 weight loss. Around two third displayed hemoptysis, dyspnea, sweat at night, and long lasting 123 sub fever at night. The cases had many symptoms being 7.5 ± 2.3 complaints/person in 124 average. These cases were registered, hence all of them underwent to treatment, whether they 125 knew it was under DOTS or not (Table 3). Nine had no supervisors and at least one fourth had 126 to pay treatment fee, suggesting that they were not under DOTS. More than 85 % of the cases 127 quit taking medicine at 6 months, regardless of frequency of medication in first 2 months. 128 Treatment with 4 drugs was the most common, but that with 2 drugs was also observed in 129 around 15 % of the cases. Sputum smear was the most common examinations during 130 treatment followed by chest X-ray. 131 The controls were selected from people visited the same health center and 132 diagnosed not having TB. We tried to obtain age-gender-matched control. We could achieve 133 gender-matched, but their age was slightly younger than the cases (Table 1). Around 80 % of 134 the cases were farmers and around 30 % of them graduated from elementary school alone. 135 More than half of the controls received BCG whereas more than half of the cases did not 136 know whether they received BCG or not. Nine percent of the cases experienced close contact 137 with TB patients but none of the controls did. Income of the cases was significantly lower 138 than that of the controls, and three fourth of them got less than 100 $/month. The cases were 139 diagnosed already, hence they got the TB information from medical staff more than the 140 controls, but they usually did not use other sources (Table 1). When the living conditions 141 were compared, all conditions were significantly different between the cases and the controls 142 (Table 4). "Ceramic floor", "outside kitchen", "gas for cooking", "open windows everyday", 143 "sunlight into the house", and "ventilation in every room" were more common in the controls, 144 whereas "window in each room" and high "humidity in the house" were in the cases. 145 The cases less frequently "washed their hands before eating" but more frequently 146 "shared the dishes with others" and "drunk from the same glasses/bottles" than the controls 147 (Table 5). They also less frequently "washed their hands after blowing" than the controls, 148 whereas no difference was observed in the frequency of whether "they worked when they felt 149 unwell" between these two groups. 150 In order to clarify what kinds of these physical factors were most affected 151 difference in the cases and the controls, the MLR was applied ( Table 6). The cases/controls 152 were the dependent variables and the groups divided by above mentioned information were 153 used as determinants. The way of obtaining TB information was removed from the 154 determinant because that from the cases was modified as described above. The obtained risks 155 were "farmers"，"close contact with TB patients", "whether or not they did not know they 156 received BCG", "smoking", and "low income; ˂ 100 $/month". "High income; ˃ 150 $/month" 157 was also extracted as a risk compared with "middle income; 100-150 $/month". To "wash 158 hands before eating", "wash hands after blow" and "not work when unwell" were protective. 159 Among living conditions, "live with ≥ 3 person in the same room", "soil floor", no "sunlight 160 in the house" and no "ventilation in the house" were extracted as risk. 161 Regardless of many symptoms, the cases did not always display significant 162 differences with the controls among the opinion regarding what they thought was TB 163 symptoms ( Figure 1). "Long lasting sub fever" alone was significantly higher in the controls. 164 More than 70 % of both the cases and the controls thought that it was important to 165 "cover mouth/nose when someone sneezed" (Figure 2). More than half of the controls thought 166 that "avoid sharing dish", "avoid drinking from the same glass/bottle", "wash hands after 167 touching items in the public" and "maintain good nutrition" were the way to avoid getting TB 168 and the rates were significantly higher than the cases. In fact, 76 % of the controls washed 169 hands before eating and only 35 % of the cases did so (Table 5). Moreover, 76 % (9 + 67) of 170 the cases at least sometimes "shared the dish" and" drunk from the same glass", that was 171 significantly higher than the controls. On the other hand, 87 % (39 + 48) of the controls at 172 least sometimes "washed hands after blowing". "Vaccination" was also higher in the controls 173 than the cases ( Figure 2). 174 Opinions related to the seriousness and shame, did not display any significant 175 difference between the cases and the controls (Table 7). Although many of the cases and 176 controls thought TB to be "serious", they did not always think that TB was "serious at 177 workplaces" and "affected work performance". Significant difference was not observed in 178 "be ashamed of having TB" but the cases were tended to want to "hide having TB". 179 Significantly more controls thought "TB affected relationship with others" and "wanted to 180 live isolated", whereas there was no significant difference in "TB affected family 181 responsibility" against the controls. Both of the cases and controls usually tried to be good at 182 TB patients. Around 50 % of the controls believed "TB treatment was very costly" but 183 around one fourth of the cases thought so. "HIV positive people should concern about TB" 184 13 was significantly higher in the controls than in the cases. Around one fourth to one third of 185 the cases and controls believed that TB was hereditary. 186

Discussion 187
In Indonesia, regular health examination was not mandatory [6]. Therefore, after symptoms 188 became apparent, the person visited the clinic where sputum smear was not always available. 189 The cases in the present study displayed 7.5 ± 2.3 complaints/person in average, indicating 190 the delay of diagnosis [15]. Although early diagnosis and initiation of treatment of infectious 191 cases is the best measure to reduce transmission [3,16,17], in some countries, 20 % of 192 patients were not diagnosed for over 6 months from the onset of symptoms [18]. Even after 193 the symptoms became obvious, for diagnosis, it took at least 2 more days because positive TB 194 was defined as more than 2 positive sputum smears in the smear performed three times within 195 2 days [14]. Household contacts continued meantime, when patients were with potentially 196 infectious forms proceeding to high prevalence of TB [19,20]. On the other hand, culture was 197 not common, whereas patients with smear-negative, culture-positive TB were reportedly 198 responsible for TB transmission [21,22]. Immediate introduction of culture examination is 199 required because, in addition to high sensitivity, it allows determining whether the patient is The cases did not always know whether they were under DOTS treatment or not, but all the 202 cases could luckily quit taking medicine regardless of the obvious delay of diagnosis. Around 203 80 % of the cases were farmer, and around 30 % graduated elementary school alone, hence 204 their income was lower than the controls. Low income and low education are reportedly 205 associated with TB infection [4,[23][24][25]. They also had lost the chance of BCG injection. This 206 occupation was also extracted as a risk by the MLR. However, the MLR extracted high 207 income as a risk as well. Such a result is not always in accordance with several studies [4, 208 23-25], whereas it is conceivable that people with high income, regardless of their occupation, 209 had more chances to live and/or work at the places with a lot of people where a risk of TB 210 transmission was supposed to be high. The number was small but only cases had a chance to 211 close contact with TB patients. TB contact was absolutely the risk of TB transmission [18,26]. 212 No significant difference in the rate of "smoking" by the MA, but the MLR extracted 213 "smoking" as a risk to be the case. This is in good accordance with the previous report [23]. It 214 is natural that the cases "source of TB information" was medical staff, but the cases were not 215 always eager to collect information from other sources comparing with the controls. All the 216 items related living condition was significantly different between the cases and the controls 217 by the MA. Among them, the MRL extracted "small number of the person in the room", 218 "ceramic floor", "sunlight in the house", and "ventilation in the house" as protective. 219 Importance of good ventilation was emphasized elsewhere [3,23]. Excluding "work when 220 unwell", their attitudes displayed significant differences between the cases and the controls by 221 the MA. "Share the dish" and "drink from the same glasses/bottles" were not extracted by the 222 MLR. Instead, "work when unwell" was extracted as well as "wash hands before eating" and 223 "wash hands after blow". These findings may be a reflection that TB is airborne. In general, 224 the cases were not aware of danger in their attitudes, which was in good accordance with 225 previous reports [10][11][12]. 226 Both of the cases and the controls did not recognized "dyspnea" and "chest pain" as TB 227 symptoms. Significant differences existed, but "long lasting sub fever" was also not 228 considered as TB symptoms. "dyspnea", "chest pain" and "long lasting sub fever" were less 229 frequent than "cough with sputum", "malaise" and so on but number of the cases complaint 230 them. Thus, it seems necessary to let the people know the TB symptoms [10][11][12]. 231 TB itself was recognized to be dangerous both by the cases and the controls, but they did not 232 recognize its dangerousness at work places. Many of the cases were farmers; hence it seems 233 less possible to spread TB than workers. However, TB positive workers can work and be able 234 to transmit TB to their colleagues. Comparing with the rate of both the cases and the controls 235 who thought "having TB was a shame", that of "wanted to hide having TB" was less. It seems 236 natural that more controls who did not receive TB treatment believed that TB was "affected 237 relationship with others" and wanted to "live isolated in case of TB" than the cases. DOTS 238 performed under the governmental hospitals and health centers were free [9], but some cases 239 visiting private hospitals/clinics had to pay the treatment fee. Higher rate of "HIV positive 240 people should be concerned about TB" in the control was reflection that they were more eager 241 to collect information than the cases. HIV infection reportedly affected TB infection [27]. 242 However, the number who believed "TB was hereditary" was not different between two 243

groups. 244
Some aspects underlying the low coverage CDR are problems of socioeconomic, 245 education/knowledge and stigma [4]. Economic conditions will affect the public in getting not 246 only good environmental home conditions but also the excellent level of education. The level 247 of education in this study was relatively low because many people only finished elementary 248 school where sufficient TB education was impossible. Poor education will cause shortage of 249 knowledge about TB, leading the public into embarrassment and sometimes attitude to hide 250 their disease if they exposed to TB. Such conditions may cause the delay for some people to 251 go to the health service [18]. As a result, TB was spread among the farmers even when their 252 contact was not always intense like workers. It is very difficult to change occupation, income 253 and housing condition, immediately. Thus, community based TB education is very important. 254 It may be useful to educate and expose not only public but also private practitioners to the 255 community based TB program [9,28]. 256 Some cases were not dependent on the free DOTS program. To inform the existence of this 257 program is also a good education. Utilization of this program not only reduces multi-drug 258 resistant TB, but also helps reducing out-of-pocket expenses to patients [27]. Number of 259 syndromes of the cases absolutely indicated the delay of diagnosis. 260 The classic symptoms of TB are fever, cough and weight loss, but they are non-specific and 261 can be mimicked by other conditions, including malignancy and other pulmonary infections. 262 That is, in an early stage, such syndromes are not always specific to TB. However, 263 importance of these classic lung related syndromes should be aware that they are possible 264 signs of initiation of TB expansion [29]. 265 266

Acknowledgements 267
We would like to thank to Semarang State University, Directorate General of Higher 268 Education (DGHE or DIKTI) and Health Department Semarang District, for all support. 269

Conflict of Interests 270
The authors declare that there is no conflict of interests regarding the publication of this 271 paper. 272
Davies  Differences in the knowledge about TB symptoms between the cases and controls (multiple answer). *Significant differences (ρ< 0.05, χ 2 test and Student's t-test).

Fig. 2
Differences in the way that the cases and controls thought it good to avoid getting TB *Significant differences (ρ < 0.05, χ 2 test). 1