INTRODUCTION
According to the World Health Organization (WHO), there were 10 million new cases of tuberculosis globally in 2017.(1) In South Africa, tuberculosis had an incidence rate of about 500,000 cases in 2017 and accounted for 6.5% of all deaths in 2016.(2) Several different types of blood abnormalities have been described in patients with both pulmonary tuberculosis (PTB) and extra-pulmonary tuberculosis (ETB) with the more common haematological abnormalities associated with PTB described as a reactive thrombocytosis and a leucocytosis.(3,4)
ETB can prove to be diagnostically challenging, especially when confined to a site such as the bone marrow or abdomen. A publication in 1943 described a spectrum of changes, including monocytosis, basophilia, leucocytosis, leucopenia, anaemia, “leukemoid reactions” and pancytopenia.(5) Since then, studies have confirmed or disputed these findings.(3,4,6–8) Pancytopenia has been reported by some authors as an unlikely manifestation of tuberculosis.(9) Some haematological abnormalities in tuberculosis also have patient outcome implications, such as presence of lymphopenia has been found to be an independent predictor of mortality in PTB.(10) Since most studies on the haematological effect of tuberculosis were performed in the pre-HIV (human immunodeficiency virus) era, there is still uncertainty about the haematological changes present in patients with tuberculosis co-infected with HIV. With an estimated overall HIV prevalence of 13.1% in 2018 in South Africa, there is a need to re-evaluate the haematological manifestations of tuberculosis in the South African population.(11)
The aim of this study was to describe the changes in full blood count in patients with tuberculosis in an evolving HIV era.
METHODS
Patients admitted to Chris Hani Baragwanath Academic Hospital with microbiologically or histologically proven confirmed tuberculosis between October 2017 and July 2018 were prospectively and consecutively recruited for the study. A list of hospitalised patients that tested positive for tuberculosis was obtained from the National Health Laboratory Service (NHLS) laboratory. This included patients with (1) positive Xpert® MTB/RIF (GXP) on sputum or pleural fluid, fine needle aspirate (FNA) of lymph node, ascitic fluid or cerobrospinal fluid (CSF); (2) positive auramine stain of sputum or pleural fluid; (3) positive Ziehl–Neelsen on FNA of lymph node; (4) positive culture of Mycobacterium tuberculosis on sputum, pleural fluid, FNA of lymph node, CSF, blood and bone marrow aspirate; (5) histological diagnosis of tuberculosis on tissue, lymph node or trephine biopsy. Exclusion criteria included pregnancy, established chronic kidney disease not related to tuberculosis, established chronic liver disease not related to tuberculosis, haemoglobinopathies and malaria. Written informed consent was obtained from all participants included in the study. This study was approved by the Human Research Ethics Committee (Medical) of the University of the Witwatersrand (M170633).
Procedures
Demographics and relevant medical history were obtained for the all study participants. The NHLS database was accessed to obtain their laboratory results. The following were collected: full blood count, differential count, reticulocyte production index, international normalised ratio, activated partial thromboplastin time, iron, ferritin, transferrin, B12, folate, electrolytes, urea, creatinine and calcium. If performed, bone marrow aspirate and trephine results were also collected.
Definitions
The modified WHO grading system for blood parameters was used. Anaemia: Hb <11 g/dl; polycythemia: Hb >18.5 g/dl ± HCT >0.52 l/l (males), Hb >16.5 g/dl ± HCT >0.48 l/l (females); leucopenia: white cell count (WCC) <4 × 109/l; leucocytosis: WCC >11 × 109/l, neutropenia: neutrophil count <2 × 109/l; neutrophilia: neutrophil count >7.5 × 109/l, lymphopenia: lymphocyte count <1 × 109/l; lymphocytosis: lymphocyte count >4 × 109/l, monocytosis: monocyte count >0.8 × 109/l; monocytopenia: monocyte count <0.2 × 109/l; clinical thrombocytopenia: platelet (Plt) count <100 × 109/l; thrombocytosis: Plt count >450 × 109/l.
Statistical analysis
Data was captured using Microsoft Excel® and transferred onto Stata® 12.0 for statistical analysis. Age and other non-normally distributed variables were presented using medians and interquartile ranges. Normally distributed parameters such as Hb were presented using means and standard deviations. Continuous data was compared using the t-test if normally distributed, for example Hb, or the Mann–Whitney test if not normally distributed as in the case of WCC. Comparison of categorical data such as leucopenia was performed using the Chi-square test. A p-value of <0.05 was considered as being statistically significant.
RESULTS
Data on 125 patients were collected over a period of 10 months. The baseline characteristics of the study population are illustrated in Table 1. About 90% of patients were in the age group 18–60 years. PTB was diagnosed through sputum GXP or pleural fluid GXP. ETB was diagnosed on lymph node GXP, CSF GXP, ascitic fluid GXP, tissue or bone marrow trephine.
Baseline characteristics of study population (n = 125) stratified according to site of tuberculosis
| Characteristic | n | PTB | ETB |
|---|---|---|---|
| Participants | 125 | 89 (71.2%) | 36 (28.8%) |
| Gender – Male | 77 | 57 (64.0%) | 20 (55.6%) |
| ▒▒Female | 48 | 32 (36.0%) | 16 (44.4%) |
| Age* | 37 (30–47) | 38 (31–50) | 35 (25–39) |
| HIV-seropositive | 101 (80.8%) | 68 (76.4%) | 33 (91.7%) |
| CD4 count* | 100 | 74 (17–188) | 37 (14–75) |
| VL* | 86 | 124,450 (277–587,000) | 75,100 (12,100–709,000) |
| On combination antiretroviral therapy (cART) | 37 | 27 | 10 |
| Method of diagnosis: | |||
| Sputum GXP | 88 | ||
| Lymph node aspirate GXP | 15 | ||
| CSF GXP | 12 | ||
| Tissue biopsy | 3 | ||
| Pleural fluid GXP | 5 | ||
| Bone marrow trephine | 4 | ||
| AGXP | 1 |
IQR = interquartile range, PTB = pulmonary tuberculosis, ETB = extra-pulmonary tuberculosis, HIV = human immunodeficiency virus, VL = HIV viral load, cART = combination antiretroviral therapy, GXP = Xpert MTB/Rif assay, AGXP = Xpert MTB/Rif assay on peritoneal fluid.
*Median + IQR.
PTB vs ETB
Most patients with anaemia were normocytic (68%) and either hypochromic (57.3%) or normochromic (41.3%). Tables 2 and 3 illustrates the differences in the haematological profile between the PTB and ETB groups. Patients with PTB had a significantly higher WCC than with ETB (7.96 (5.27–10.34) vs 5.59 (3.6–8.1) p = 0.002). Leucocytosis was only seen in patients with PTB, mainly due to high neutrophil counts. Neutrophils were found to be higher among PTB patients bordering on statistical significance (5.5 (3.3–8.4) vs 4.2 (2.1–7.0), p = 0.079), but these were still within the normal range. One patient in the PTB group had a platelet count above 1000 × 109/l. Patients with tuberculous meningitis showed no cytopaenias besides anaemia.
Comparison of the proportion of haematological abnormalities of patients with PTB and patients with ETB (PTB vs ETB)
| Total population | PTB | ETB | p Value | HIV+ | HIV− | p Value | |
|---|---|---|---|---|---|---|---|
| N | 125 | 89 | 36 | 101 | 24 | ||
| Anaemia* | 76 (60.8%) | 54 (60.7%) | 22 (61.1%) | 0.964 | 67 (67.3%) | 9 (37.5%) | 0.010 |
| Polycythemia | 1 | 0 | 1 | 1 (1%) | |||
| Leucopenia* | 25 (20%) | 13 (14.6%) | 12 (30%) | 0.018 | 24 (23.8%) | 0 | |
| Leucocytosis | 19 (15.2%) | 19 (21.3%) | 0 | 0.003 | 13 (12.9%) | 10 (41.7%) | 0.137 |
| Thrombocytopenia | 15 (12%) | 10 (11.2%) | 5 (13.9%) | 0.679 | 12 (11.9%) | 3 (12.5%) | 0.933 |
| Thrombocytosis | 17 (13.6%) | 13 (14.6%) | 4 (11.1%) | 0.505 | 10 (9.9%) | 0 | |
| N | 76 | 52 | 24 | 63 | 13 | ||
| Neutropenia* | 11 (14.5%) | 6 (11.5%) | 5 (13.9%) | 0.284 | 11 (17.4%) | 0 | |
| Neutrophilia | 22 (28.9%) | 17 (32.7%) | 5 (20.8%) | 0.224 | 16 (25.4%) | 6 (46.2%) | 0.042 |
| N | 75 | 52 | 23 | 62 | 13 | ||
| Lymphopenia | 48 (64%) | 31 (59.6%) | 17 (73.9%) | 0.346 | 45 (72.6%) | 3 (23.1%) | 0.001 |
| N | 76 | 52 | 24 | 63 | 13 | ||
| Monocytopenia | 14 (18.4%) | 9 (17.3%) | 5 (20.8%) | 0.712 | 14 (22.2%) | 0 | |
| Monocytosis | 11 (14.5%) | 9 (17.3%) | 2 (8.3%) | 3 (4.8%) | 2 (%) | 0.160 | |
| Anaemia only** | 54 (43.2%) | 41 (46.1%) | 13 (36.1%) | 45 (44.6%) | 9 (37.5%) | ||
| Bicytopenia | 17 (13.6%) | 12 (13.5%) | 5 (13.9%) | 16 (15.8%) | 1 (4.2%) | ||
| Pancytopenia | 6 (4.8%) | 2 (2.2%) | 4 (11.1%) | 6 (5.9%) | 0 |
HIV+ = HIV-seropositive, HIV− = HIV-seronegative, PTB = pulmonary tuberculosis, ETB = extra-pulmonary tuberculosis.
*Grade 1–4.
**From grade 1.
Haematological parameters of the sample population stratified according to site of tuberculosis (PTB vs ETB)
| n | PTB | ETB | p Value | |
|---|---|---|---|---|
| WCC (×109/l)* | 125 | 7.96 (5.27–10.34) | 5.59 (3.58–8.1) | 0.002 |
| RCC (×1012/l)** | 125 | 3.577 ± 0.88 | 3.82 ± 0.98 | 0.789 |
| Hb (g/dl)** | 125 | 10.3 ± 2.7 | 10.2 ± 2.7 | 0.816 |
| HCT (l/l)** | 125 | 0.324 ± 0.77 | 0.324 ± 0.084 | 0.996 |
| MCV (fl)* | 125 | 86.0 (80.9–90.3) | 85.6 (76.3–93.6) | 0.840 |
| Plt (×109/l)* | 125 | 262 (144–393) | 262 (189–402) | 0.817 |
| Neutrophils (×109/l)* | 76 | 5.5 (3.3–8.4) | 4.2 (2.1–7.0) | 0.079 |
| Lymphocytes (×109/l)* | 76 | 0.82 (0.49–1.49) | 0.73 (0.42–1.13) | 0.365 |
| Monocytes (×109/l)* | 76 | 0.45 (0.27–0.78) | 0.37 (0.21–0.58) | 0.213 |
| Eosinophils (×109/l)* | 29 | 0.05 (0.01–0.13) | 0.02 (0–0.1) | 0.287 |
PTB = pulmonary tuberculosis, ETB = extra-pulmonary tuberculosis, WCC = white cell count, RCC = red cell count, Hb = haemoglobin, HCT = haematocrit, MCV = mean corpuscular volume, Plt = platelet count.
*Median + IQR.
**Mean ± standard deviation.
PTB vs ETB with HIV infection
Our sample consisted of 80.8% seropositive patients, 76.4% with PTB and 91.6% with ETB. Among our HIV-seropositive patients (76.4%), we observed anaemia in 60.8% of patients, 67.6% in the PTB and 63.6% in the ETB group, but this difference was not statistically significant. WCC was again statistically significantly higher in the PTB HIV+ group 6.77 (4.8–9.7) vs 5.27 (3.39–8.77), p = 0.033. Total protein and serum albumin levels were lower among patients with PTB HIV+, but the difference was only statistically significant for protein levels ((71.6 ± 12.8) vs. (77.6 ± 11.1), p = 0.035) and ((25.9 ± 6.4) vs. (28.5 ± 7.4), p = 0.099), respectively. No differences in neutrophil count, lymphocyte count, platelet count or electrolytes were observed between these two groups.
PTB vs ETB HIV-negative patients
In HIV-seronegative patients, the only abnormalities were a higher WCC and neutrophil count in the PTB group (9.9 (8.6–11.4) vs 5.8 (5.7–7.2), p = 0.026) and (8.51 (5.85–10.13) vs 3.75 (3.05–4.45), p = 0.030). All other measured parameters were similarly distributed between the HIV-seronegative PTB and ETB groups.
PTB vs ETB in patients on cART
Among HIV-seropositive patients on cART, WCC was similarly distributed between the PTB and ETB groups. The following parameters were lower among patients with ETB on cART compared to patients with PTB on cART: Hb (9.6 ± 2.3) vs (11.3 ± 2.3), p = 0.043, albumin (25.4 ± 7.4) vs (32.9 ± 8.9), p = 0.020, protein (71.4 ± 11.0) vs (80.1 ± 5.2), p = 0.310, aspartate transaminase (56 (30–91) vs 27 (24–40), p = 0.049), total bilirubin (9 (6–17) vs 6 (4–6), p = 0.071), conjugated bilirubin (5 (4–12) vs 3 (2–3), p = 0.002). The rest of the blood parameters were not different between the two groups.
PTB not on cART vs ETB not on cART
In the 64 HIV-seropositive patients not on cART, we observed statistically significantly higher levels of WCC in the PTB group (6.1 (4.8–8.4) vs 4.8 (3.0–7.3), p = 0.043). All other parameters were similarly distributed between these two groups.
Effects of cART
Patients on cART (combined PTB and ETB) had higher levels of WCC and platelet count when compared to patients not on cART (7.9 (4.6–10.9) vs 5.8 (4.3–8.3), p = 0.047) and (316 (217–436) vs 233 (119–318), p = 0.020), respectively. The effect of cART was then assessed among patients with PTB and ETB separately. No statistically significant haematological changes were noted between the PTB not on cART group and PTB on cART group. Of note, PTB patients on cART demonstrated lymphopenia in 73.3%. The significant differences attributable to cART in patients with ETB were a higher haemoglobin (11.3 ± 2.3) vs (9.3 ± 2.2), p = 0.026, and higher albumin levels (32.9 ± 8.9) vs (26.4 ± 5.8), p = 0.028.
Effects of HIV infection
Table 4 illustrates the significant differences in blood parameters between HIV-seropositive and HIV-seronegative patients. Hypoalbuminaemia was present in 80.8% of our study population, being present in 88.1% of HIV-seropositive patients and 50% of HIV-seronegative patients. Neutrophilia was present in 32.7% of patients with PTB and 26.8% if co-infected with HIV.
Comparison of blood parameters of all patients in study population stratified by HIV status – only parameters with significant differences shown
| Parameter | n | HIV-seropositive | HIV-seronegative | p Value |
|---|---|---|---|---|
| WCC (×109/l)* | 125 | 6.1 (4.4–9.3) | 9.4 (8.0–11.2) | 0.001 |
| RCC (×1012/l)** | 125 | 3.7 ± 0.8 | 4.1 ± 1.1 | 0.046 |
| Hb (g/dl)** | 125 | 9.9 ± 2.5 | 11.5 ± 3.0 | 0.009 |
| HCT (l/l)* | 125 | 0.31 (0.26–0.36) | 0.37 (0.31–0.42) | 0.005 |
| Neutrophils (×109/l)* | 76 | 4.7 (2.5–7.6) | 7.5 (5.1–9.4) | 0.014 |
| Lymphocytes (×109/l)* | 76 | 0.69 (0.39–1.18) | 1.51 (1.13–1.93) | 0.001 |
| Monocytes (×109/l)* | 76 | 0.38 (0.21–0.59) | 0.77 (0.48–0.89) | 0.005 |
| Bicarbonate (mmol/l)* | 125 | 20 (17–21) | 20.5 (18.5–23.5) | 0.027 |
| Calcium (mmol/l)** | 85 | 2.04 ± 0.19 | 2.15 ± 0.17 | 0.046 |
| Albumin (g/l)** | 104 | 26.7 ± 6.8 | 33.0 ± 8.5 | 0.001 |
| AST (U/l)* | 104 | 53.5 (31–87) | 21.5 (17–53) | 0.001 |
WCC = white cell count, RCC = red cell count, Hb = haemoglobin, HCT = haematocrit, MCV = mean corpuscular volume, AST = aspartate transaminase.
*Median + IQR.
**Mean ± standard deviation.
PTB and the effect of HIV infection
In Table 5, we can observe the significant differences attributable to HIV on the patients diagnosed with PTB. The PTB HIV+ group had significantly lower WCCs, haemoglobin, neutrophils, lymphocytes, monocytes and albumin. However, despite these parameters being lower in the PTB HIV+ group, they remained within the normal range. Thirty-six patients (28.8%) were diagnosed with ETB, most of whom (91.7%) were HIV-seropositive. Lymphopenia was common among HIV-seropositive patients in this group at 73.9%.
Blood parameters of patients with PTB stratified by HIV status – only parameters with significant differences shown (PTB HIV+ vs PTB HIV−)
| Parameter | n | HIV-seropositive | HIV-seronegative | p Value |
|---|---|---|---|---|
| WCC (×109/l)* | 89 | 6.9 (4.8–9.7) | 9.9 (8.6–11.4) | 0.004 |
| Hb (g/dl)** | 89 | 9.9 ± 2.6 | 11.4 ± 2.7 | 0.030 |
| HCT (l/l)* | 89 | 0.305 (0.257–0.358) | 0.359 (0.321–0.417) | 0.007 |
| Neutrophils (×109/l)* | 52 | 4.7 (2.8–7.7) | 8.5 (5.9–10.1) | 0.011 |
| Lymphocytes (×109/l)* | 52 | 0.7 (0.46–1.31) | 1.51 (1.13–2.05) | 0.005 |
| Monocytes (×109/l)* | 52 | 0.39 (0.24–0.56) | 0.80 (0.48–0.98) | 0.006 |
| ALT* | 73 | 31 (14.5–53) | 20 (11–33) | 0.058 |
| AST* | 73 | 64 (34.5–96.5) | 22 (18–65) | 0.006 |
| GGT* | 73 | 77.5 (47–156) | 54 (35–84) | 0.055 |
| Albumin* | 73 | 25.5 (22–30) | 32 (28–41) | 0.002 |
WCC = white cell count, RCC = red cell count, Hb = haemoglobin, HCT = haematocrit, MCV = mean corpuscular volume, ALT = alanine transaminase, AST = aspartate transaminase, GGT = gamma glutamyltransferase.
*Median + IQR.
**Mean ± standard deviation.
ETB and the effect of HIV infection
Lymphopenia was observed in 80% of patients in the ETB HIV+ group. Levels of lymphocytes in the upper ranges of normal were seen in patients with ETB that were HIV-seronegative. We did not observe any difference between the results of the ETB HIV+ vs ETB HIV− groups.
Bone marrow aspirates and trephine
Only six patients had a bone marrow aspirate and trephine biopsy performed. Four of the patients had a diagnosis of PTB and the remaining two, ETB, based on the method of initial diagnosis. Four patients had a bicytopenia and two patients had a pancytopenia. Overall, no features of tuberculosis were observed on any bone marrow aspirates. Plasmacytosis was not seen on any of the samples. Four bone marrow trephine specimens had evidence of tuberculosis with granulomata and acid-fast bacilli observed in two. All these patients either had a pancytopenia or a bicytopenia in their peripheral blood. All the patients exhibited a peripheral lymphopenia and half of them had a blood monocytopenia. The presence of bone marrow granulomata was associated with thrombocytopenia in all cases. There was no evidence of a neoplastic infiltrate in any of the trephine biopsies.
DISCUSSION
To our knowledge this is the first South African study to highlight the changes in full blood count of patients with tuberculosis in the HIV era. Anaemia is still the most common manifestation irrespective of mode of diagnosis, HIV status or use of antiretroviral therapy. There was no statistically significant difference in the degree of anaemia in patients with PTB and ETB. At 60.7% in the PTB group, our reported frequency of anaemia was similar to a study in the pre-HIV era by Morris et al.,(4) but much lower than the 76% reported by Baynes et al.(12) It is noteworthy that these studies used different definitions of anaemia. While Morris et al. used Hb <13 g/dl for males and <11 g/dl for females, Baynes et al. used Hb <14.3 g/dl in males and Hb <12.1 g/dl for females.(4,12) Among the ETB patients, anaemia occurred in 61.1%, which is the approximate average value of previously published studies. Lombard and Mansvelt reported the prevalence of anaemia at 72% in their cohort of patients with miliary tuberculosis and Maartens et al. observed anaemia in 52% of their patients.(3,13) Similar to Amilo et al., anaemia was significantly more severe in our study patients co-infected with HIV.(14)
The mechanism of anaemia is probably similar to the anaemia of other chronic disorders. Tuberculosis, through the release of IL-6, stimulates hepcidin synthesis which in turn causes internalisation and degradation of ferroportin in reticuloendothelial cells and duodenal enterocytes.(15) The net effect is reduced iron availability for erythropoiesis. On the other hand, most of our patients exhibited high levels of ferritin, and low iron level which is thought to be related to IL-1 activation by monocytes.
In our study population, PTB was associated with significantly higher WCCs than ETB. The PTB group on cART had similar WCCs to the PTB HIV-seronegative group. This suggests that antiretroviral therapy restores the body's ability to mount an immune response to tuberculosis.
Overall, co-infection with HIV was associated with lower levels of WCC across all PTB groups. This is in agreement with other studies in the literature.(14,16) This could explain why 59.6% of our PTB patients had lymphopenia as our cohort consisted of a majority of HIV-seropositive patients. Lymphopenia was observed in only 18.1% of HIV-seronegative patients. Lymphopenia has classically been associated with ETB and not PTB. In the subset PTB HIV+, 70.7% of patients had lymphopenia demonstrating the compounding effect of HIV on lymphopenia in patients with PTB. Interestingly, even PTB patients on cART demonstrated lymphopenia at 73.3%.
Lymphopenia was highly prevalent among ETB patients (73.9% of all ETB patients and 80% in the ETB HIV+ group). Levels of lymphocytes in the upper ranges of normal were seen in patients with ETB who were HIV-seronegative. In comparison, Maartens et al. reported lymphopenia in 87% of their patients with milliary tuberculosis using a cut-off of <1.5 × 109/l.(3) Lymphopenia in tuberculosis is thought to be a reaction to the mycobacterial infection and not due to an underlying immunodeficiency.(17) T4 lymphocytes are recruited at the areas of granuloma formation.(18) Lymphopenia is of clinical relevance as it has been established as an independent predictor of mortality in patients with PTB.(10)
It appears that cART also has some effects on neutrophils. In the current study neutrophilia was present in 32.7% of patients with PTB and 26.8% if co-infected with HIV. This is lower than the 40% reported by Morris et al.(4) The average neutrophil counts of PTB patients was lower in the HIV-seropositive group than in the HIV-seronegative group. Higher levels of neutrophils were present among patients on cART than those were not on cART, but this was not statistically significant. It seems that HIV impairs the neutrophilic response to tuberculosis and cART is useful in its restoration.
All of our six patients with evidence of tuberculosis on bone marrow were HIV-seropositive. While they all exhibited lymphopenia, half of them demonstrated monocytopenia. These results are in accordance with those of Lombard and Mansvelt who described lymphopenia and monocytopenia in their 25 HIV-seronegative patients.(13) Cucin et al. reported monocytosis in 22% of their patients with miliary tuberculosis.(19) In contrast, monocytosis was more common with PTB than ETB (17.8% vs 8.3%) in our study. In comparison to a study by Morris et al. that demonstrated plasmacytosis in some cases, none of our patients had this finding, but this may be due to the small number of bone marrow aspirates performed.(4)
Thrombocytosis has been reported as a haematological manifestation of PTB. Our cohort consisted of 15.7% of PTB patients with thrombocytosis with one patient having a platelet count above 1000 × 109/l. Thrombocytosis was also observed by Baynes et al. and was reported at a similar proportion (12.5%) by Kutiyal et al.(20,21)
No significant differences were seen in platelet counts in the group comparisons. Clinical thrombocytopenia was relatively uncommon in our sample but was uniformly present in all four patients who had bone marrow infiltration with granulomata. Cucin et al. and Maartens et al. have previously reported the correlation between thrombocytopenia and the presence of granulomata in the bone marrow.(3,19)
The absence of cytopenias among patients with tuberculous meningitis suggests that tuberculous meningitis is more of a localised disease. The absence of any thrombotic manifestations in our sample was unexpected since tuberculosis and HIV have both been described as pro-coagulable states.(22,23)
Limitations
The strength of this single centre study was that it was prospective; however, the total sample size was 125 patients, and this meant that all sub-group analyses performed were on much smaller sample groups. While some sub-groups such as the PTB probably had numbers large enough for relevant analysis, others, such as the ETB sub-group, did not. We could not establish the effect of co-trimoxazole prophylaxis on our patients since a very small number were on this therapy.
CONCLUSION
Anaemia is still the most common manifestation of tuberculosis. We have highlighted the reactive nature of PTB haematological abnormalities, as compared to ETB based on the presence of higher WCCs in both HIV-seropositive and HIV-seronegative patients. Co-infection with HIV was associated with lower levels of haemoglobin, WCC, neutrophil count and lymphocyte count among patients with PTB. The end result is that HIV seems to blunt the reactive response to PTB to some extent. Further studies are required to evaluate the effect of HIV on ETB.