Treatment outcomes and risk factors of extra-pulmonary tuberculosis in patients with co-morbidities

A retrospective study was conducted in four states of Malaysia from 2006 to 2008. Data on socio-demographic, clinical, histopathological, microbiological and other laboratory variables of EPTB cases were collected from TB registers, treatment cards and TB medical personal files using standard data collection tool. Patients in whom the site of infection was confined to lungs were considered as PTB while infection extended to other organs or tissues outside lungs were considered as EPTB. Patients who had both PTB and EPTB involvements were excluded from the analysis of EPTB based on WHO sample selection policy [11]. The sites of infection were as lymph nodes, gastrointestinal system, spinal, meningitis pleural effusion, miliary and bones/joints. All other sites of infection were considered as part of a seventh group identified as “other rare forms” (including urogenital, eyes, ear, breast etc.) for convenient statistical analysis the diagnosis of EPTB was done following Clinical Practice Guidelines [12], which is consistent with the WHO’s diagnostic criteria [13]. Treatment success refers to the patients who were cured and have completed TB treatment. Completed TB treatment was defined as any patient who had completed a TB regime based on the Clinical Practice Guidelines created by Malaysia Ministry of Health [12].

The study was carried out in selected hospitals and prisons located in four states of Malaysia (Penang, Sabah, Sarawak, and Selangor). The states of Penang, Sabah and Sarawak were selected for the present study based on previous literature report of TB burden. The prevalence of TB in Malaysia was highest in Sabah followed by Sarawak and Penang [6]. The state of Selangor was also considered for the present study as it has two big prisons at national level in Malaysia.

The whole data collection form was assigned a serial number to ensure the traceability. Coding of the responses was carried out and the data was entered into the computer and analysed by using statistical package for SPSS for Windows version 16.0 (SPSS, Inc., Chicago, IL, USA). Logistic regression model was used to analyze the predictors for EPTB and unsuccessful treatment outcomes. However, p value was used to calculate significance of of co-morbidities among different variables. Factors found significant in univariate analysis, were finally included in multivariate logistic regression to estimate the odds ratios (ORs) with their 95% confidence intervals (CIs). A p value < 0.05 was considered statistically significant.

Proportion of EPTB registered in four states of Malaysia during study time period is shown in Table 1. Patients with EPTB constituted about 13.1% of all TB cases, with higher prevalence in Selangor than other three states. Socio-demographic and clinical characteristics of patients are shown in Table 2. The study cohort included 778 (63.7%) males and 444 (36.3%) females. Of the 1222 cases included in the present study, age group 26–35 years constituted the highest cases (24.2%) and least numbers were recorded for ≤15 years old (4.5%). In terms of residence, higher proportions of EPTB were observed in the urban areas (67%). The results showed that there was a significant difference (p <  0.001) in the incidence of EPTB among different races. The Malays seemed to be highly affected with a number of 389 (31.8%) cases followed by the Chinese (23.9%).

Around, 330 (27%) of cases had acid fast bacilli smear positive and 687 (56.2%) culture positive. However, 360 patients (29.5%) were diagnosed via pathology alone and 43 (3.5%) were confirmed on polymerase chain reaction (PCR). Among the culture confirmed EPTB cases, 1.5% of patients were resistant to single first line drugs. On baseline 4 patients were recorded resistant to isoniazid and 4 for streptomycin whereas 1 each for rifampicin and ethambutol. Of HIV co-infection, 139 patients were receiving Highly Active Antiretroviral Therapy (HAART). For EPTB patients with DM, 152 were getting oral hypoglycemics, 24 patients were on insulin whereas 6 had oral hypoglycemic agents plus insulin.

Of 9335 all TB cases registered during the study time period, 1222 had EPTB. Lymph node 324 (26.5%) and Pleural effusion 227 (18.6%) TB were the most frequent types of all EPTB cases among study participants (Table 2). The proportions of different types of EPTB varied with statistically significant difference observed among gender (p = 0.03), age groups (p <  0.001), different ethnic groups (p <  0.001), co-morbidities (p <  0.001), and smokers (p <  0.001) (Table 2).

Out of 1222 EPTB patients, 525 (43%) were recorded with co-morbidities. Chi-square analysis of categorical variables of the study participants showed a significant difference in males and females, distribution of age, races, residence, patient categories, marital status, education and employment between EPTB and comorbidities groups (Table 3). Among the co-morbidities, HIV and DM contributed to the highest cases and almost at the equal rate. Overall, higher proportions of all comorbidities were seen among males than females and patients aged 35 years or older. Numbers of EPTB with diabetes mellitus (EPTB-DM), EPTB with human immunodeficiency virus (EPTB-HIV) and EPTB and hepatitis (EPTB-HEP) cases were frequently seen among 35–55 years age whereas the co-morbidities became more complex with increasing age. Patients from the rural areas were significantly had increased proportions of HIV related comorbidities. Moreover, patients with CD4+ lymphocyte cell counts < 100 had 37.3% lymph node TB, 18.6% cases of miliary and pleural TB each. We further confirmed that of total deaths among known CD4+ lymphocyte count, 55% occurred in patients with < 100 counts. Other diseases which patients already had at the baseline include hypertension (3%), ischemic heart disease (1%), renal failure (2%), lung carcinoma (0.2%), lung fibrosis (0.2%), liver cirrhosis (0.2%), hypertension and COPD (0.2%), ischemic heart disease+ renal failure+ hypertension (0.4%) (Table 3).

Around, 67.6% (826/1222) patients successfully completed treatment. Treatment outcomes in relation to socio-demographic characteristics and co-morbidities are shown in Table 4. There were no statistically significant differences seen with regard to treatment outcomes among males and females, residency, ethnicity, alcohol habit and employment on univariate analysis (Table 4).

On multivariate analysis (Table 4), age group 56–65 years (1.658 [1.157–2.376]; p = 0.006), relapse cases (7.078 [1.585–31.613]; p = 0.010), EPTB-DM (1.773 [1.165–2.698]; p = 0.008), patients with no formal (2.266 [1.254–4.095]; p = 0.001) and secondary level of education (1.889 [1.085–3.288]; p = 0.025) were recorded as statistically positive significant risk factors for unsuccessful treatment outcomes. Comparing the proportion of default and deaths among different types of EPTB, significantly higher were reported in meningitis and miliary TB (Fig. 1).

Of 139 patients who were on HAART, 111 (79.9%) successfully completed the treatment while the remaining 28 (20.1%) had unsuccessful treatment outcomes. Nevertheless, statistically significant association was observed between DM and treatment outcomes for EPTB-DM patients. When death and default cases were compared among the different co-morbidities, maximum death cases were observed for EPTB-DM-HEP followed by EPTB-DM-HIV. However, higher proportions of default were seen among patients with EPTB-DM (Fig. 2).

With EPTB as the case group and PTB as the control group, we assessed the associations between the variables and having EPTB based on logistic regression model. Based on the result of the adjusted odd ratio (AOR), females appeared to have a higher risk for having EPTB (1.524 [CI: 1.311–1.746]; p <  0.001) than males (Table 5). Malays (1.251 [1.056–1.482]; p = 0.010), Indians (1.450 [1.142–1.842]; p = 0.002), urban residents (1.272 [1.109–1.459]; p = 0.001), patients with no formal education (1.361 [1.018–1.820]; p = 0.037), those with married (1.199 [1.038–1.384]; p = 0.014) and with unknown marital status (2.757 [1.611–4.717]; p <  0.001) had significantly higher odds for having EPTB. Among the co-morbid conditions, patients with EPTB-HIV (3.215 [2.347–4.405]; p <  0.001), EPTB-DM-HIV (4.361 [1.657–11.474]; p = 0.003) and EPTB-HIV-HEP (4.083 [2.785–5.987]; p <  0.001) found to have increased risk of EPTB.