- Email: [email protected]
Poor performance status is associated with early death in patients with pulmonary tuberculosis
Transactions of the Royal Society of Tropical Medicine and Hygiene (2006) 100, 681—686
available at www.sciencedirect.com
journal homepage: www.elsevierhealth.com/journals/trst
Poor performance status is associated with early death in patients with pulmonary tuberculosis S. de Valli` ere a,∗, R.D. Barker b a b
Pietersburg-Mankweng Hospital Complex, Private Bag X9316, Polokwane 0700, South Africa Department of Respiratory Medicine, King’s College Hospital, Camberwell, London, UK
Received 6 May 2005 ; received in revised form 5 September 2005; accepted 5 September 2005 Available online 28 November 2005
KEYWORDS Pulmonary tuberculosis; Performance status; Early death; South Africa
Summary The objective of this study was to determine whether poor performance status at the start of anti-tuberculous (anti-TB) treatment is associated with early death in patients admitted to hospital with pulmonary tuberculosis (PTB). During 3 months in 2001, all adult patients admitted to eight hospitals in Limpopo Province, South Africa, and diagnosed with PTB were eligible for inclusion. At initiation of anti-TB treatment, a performance status between 0 and 4 was estimated for each patient using a modified version of the Eastern Cooperative Oncology Group scoring system. Hospital records and local TB registers were reviewed to identify patients who had died during the first 2 months of treatment. In addition, it was ascertained whether a death notification had been received by the provincial administration. Fifty-three of 295 (18%) patients died within 2 months. Mortality increased from 6% in patients with the best performance status to 51% in patients with the poorest performance status. Univariate and multivariate Cox regression analysis showed that the hazard ratio for dying was significantly higher for patients with a performance status of 3 or 4. Poor performance status shows a strong association with early death in patients with PTB and has the potential to be a useful clinical, epidemiological and research tool. © 2005 Royal Society of Tropical Medicine and Hygiene. Published by Elsevier Ltd. All rights reserved.
1. Introduction Despite the availability of potent chemotherapeutic agents, patients with tuberculosis (TB) continue to die. It is
∗ Corresponding author. Present address: Division of Infectious Diseases, St. Louis University Hospital, 3635 Vista Ave, FDT-8N, St. Louis, MO 63110, USA. Tel.: +1 314 577 8648; fax: +1 314 771 3816. E-mail addresses: [email protected] (S. de Valli` ere), [email protected] (R.D. Barker).
estimated that 2 to 3 million patients die from TB every year worldwide (Dye et al., 1999). According to WHO, the mortality of new sputum-smear-positive (ss+) patients should be less than 5%, but many countries do not achieve this goal (WHO, 2005). Of the 22 countries with the highest TB incidence in the world, 10 reported a mortality of new ss+ patients of over 5% (5.4—11%). Among those high-burden countries, South Africa reported a mortality of new ss+ cases of 7%. Co-infection with HIV has been identified as a major factor contributing to increased mortality in patients with pulmonary tuberculosis (PTB) (Ashley et al., 2000). However,
0035-9203/$ — see front matter © 2005 Royal Society of Tropical Medicine and Hygiene. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.trstmh.2005.09.007
682
S. de Valli` ere, R.D. Barker
many other factors, such as delay in diagnosis, inadequacy of chemotherapeutic regimens, concomitant diseases, malnutrition and others, might affect the outcome of treatment in patients with PTB. In 2001 the Global Fund to fight AIDS, TB and malaria was created with the aim of ensuring additional resources to reduce the morbidity and mortality from these three diseases. Reduction of death from TB is also an element of the ‘Millennium Development Goal’, which has a target to halve TB-associated mortality by 2015. It would be useful to have a simple tool to identify which patients with PTB are most at risk of dying. Such a tool could be used for clinical purposes, as well as to investigate the potential benefit of new interventions that have been proposed in the past (Harries et al., 2001). The objective of this study was to determine whether the performance status of PTB patients assessed at the time of initiation of anti-TB treatment was associated with death within 2 months.
2. Materials and methods The study was carried out in the Limpopo Province, South Africa, during 2001. There were 38 hospitals in the Limpopo Province, and eight were selected for the study based on their availability of medical and nursing staff to support the study and the knowledge that they were treating a significant number of patients with PTB. For 3 months, all adults aged 20 years or older admitted to those hospitals and put on anti-TB treatment for PTB were eligible for inclusion in the study. The diagnosis of PTB was considered established if the patient was entered into the local TB register as a PTB case. Before the start of the data collection, one of the authors visited all the participating hospitals and organized a training seminar with the staff that would be involved in the data collection. At the time anti-TB treatment was started, the doctor or the nurse in charge completed the first section of a data collection sheet. This section required the recording of the patient’s name, gender, date of birth, national ID number if available, date on which anti-TB treatment was started, history of previous treatment for TB and the estimation of a performance status. For this purpose, we used a modified version of the scoring system proposed by the Eastern
Cooperative Oncology Group (ECOG) (Table 1) (Oken et al., 1982). We added to each category a brief description of the type of activity the patients should be able to carry out, taking local conditions into consideration. In all participating hospitals, the TB wards were situated on the ground level. The bathrooms were in proximity to the patient rooms (less than 10 m), and the patients had direct access to a backyard. The authors of this paper were not involved with the management of the patients. On discharge from the hospital, or if the patient died, the second section of the data collection sheet was completed, which required the recording of the sputum acid-fast bacillus (AFB) smear results as positive, negative or unknown. Sputum smear was considered unknown if the patient was unable to produce a sputum specimen that could be sent for analysis to the laboratory. The HIV test result was also recorded if available. As this was an observational study, an HIV test result was not mandatory. Finally, date of death was recorded if applicable. Three months after completion of the initial data collection, the vital status within 2 months of starting anti-TB treatment of the enrolled patients was ascertained by crosschecking hospital records and local TB registers. As a supplement, it was ascertained whether the provincial administration had received a death notification. The provincial administration was supplied with the name, first name, date of birth and the national ID number, if available, of all the patients enrolled in the study. For statistical analysis, the software package EpiInfo version 3.3.2 (CDC, Atlanta, GA, USA) was used. The 2 test and the 2 test for linear trend were used to test for differences in proportions and linear trends, respectively. Kaplan Meier survival estimates were compared using the Cox-Mantel test (log-rank test). Univariate and multivariate Cox regression analyses were performed to calculate hazard ratios with 95% CI. For the multivariate analysis, a stepwise backward model was used. The performance status was the dependent variable. The variables with a significance level >0.05 were sequentially excluded.
3. Results During the 3-month study period, 343 patients were admitted for suspected PTB to the adult wards of the eight
Table 1 Definition of the five categories of the Eastern Cooperative Oncology Group (ECOG) performance scoring system and the additional descriptions used to take local conditions into consideration ECOG categories
Additional description of performance
Score
Able to carry out normal activity without restriction Unable to do physically strenuous activity, but ambulatory and able to carry out light work Ambulatory and capable of all self-care, but unable to carry out any work; up and about <50% of waking hours Capable of only limited self-care; confined to bed or chair >50% of waking hours Completely disabled; cannot carry out any self-care; totally confined to bed or chair
No physical restriction
0
Able to walk around the whole hospital premises, but unable to do any income-generating work Able to walk around the ward and to the backyard
1
Able to go to the bathroom, but unable to walk around the ward Unable to go to the bathroom
2
3 4
Performance status in pulmonary tuberculosis
683
participating hospitals. Forty-eight patients were excluded for the following reasons: no estimation of performance status (7 patients); age missing or less than 20 years (37 patients); and anti-TB treatment started before admission (4 patients). Characteristics of the 295 included patients are shown in Table 2: 61% were male; the median age was 37 (range 20—81) years; 229 (78%) patients were ss+; HIV test result was only available for 95 (32%) patients, but 75 (79%) of the tested patients were seropositive; 16 of 20 (80%) HIV-negative patients were ss+, while only 46 of 75 (61%) HIVpositive patients were ss+ — this difference was, however, not statistically significant (2 = 2.4, P = 0.119). On the performance status scale from 0 (best condition) to 4 (worst condition), 76 of 295 patients (26%) had a performance status of 3 or 4 (Table 2). Significantly poorer performance status was seen in women, HIV-positive subjects, and patients with negative or unknown sputum smears. Overall, 53 (18%) patients died within the study period. Forty-three (15%) patients died within 1 month, while 10 (3%) patients died during the second month of anti-TB treatment. The Kaplan Meier survival curves showed an increased risk of dying with poorer performance status (P < 0.001 by log-rank test) (Figure 1). A sub-analysis of the 229 sputum AFBpositive patients showed similar Kaplan Meier survival curves (data not shown). Looking at the entire cohort, mortality
Table 2
Figure 1 Kaplan Meier survival curves according to performance status for 295 patients admitted to hospital and started on anti-tuberculous treatment; P < 0.001 by the Cox-Mantel test (log-rank test). PS: performance status.
increased from 6% in the patients with the best performance status to 51% in the patients with the poorest performance status (Table 3). Univariate Cox regression analysis showed that patients with performance status of 3 and 4 had hazard ratios of 6.2 (95% CI 2.2—17.5) and 11.1 (95% CI 4.6—27.2),
Characteristics and performance status of 295 patients admitted to hospital and started on anti-tuberculous treatment n
P-valuesa
Performance status, n (%) 0
1
2
3
4
Gender Male Female Missing
177 114 4
71 (40) 23 (20)
27 (15) 27 (24)
40 (23) 28 (25)
14 (8) 12 (11)
25 (14) 24 (21)
Age group (years) 20—34 35—49 ≥50
114 119 62
40 (35) 36 (30) 18 (29)
23 (20) 21 (18) 11 (18)
29 (25) 28 (24) 13 (21)
7 (6) 9 (8) 11 (18)
15 (13) 25 (21) 9 (15)
0.263c
HIV Positive Negative Missing
75 20 200
11 (15) 11 (55) 72 (36)
10 (13) 3 (15) 42 (21)
17 (23) 6 (30) 47 (24)
13 (17) 0 (0) 14 (7)
24 (32) 0 (0) 25 (13)
<0.001d
Sputum AFB smear Positive Negative or unknown
229 66
80 (35) 14 (21)
43 (19) 12 (18)
59 (26) 11 (17)
17 (7) 10 (15)
30 (13) 19 (29)
0.003e
Previous anti-TB treatment No 252 Yes 39 Unknown 4
76 (30) 18 (46)
48 (19) 6 (15)
60 (24) 9 (23)
26 (10) 1 (3)
42 (17) 5 (13)
National ID number Unavailable Available
116 179
45 (39) 49 (27)
23 (20) 32 (18)
21 (18) 49 (27)
10 (9) 17 (9)
17 (15) 32 (18)
Total
295
94 (32)
55 (19)
70 (24)
27 (9)
49 (17)
0.008b
0.247f
0.194g
P-values of the 2 tests comparing the distribution of the performance status of: males with females (b); of the different age groups (c); of HIV-positive with HIV-negative patients (d); of patients with positive sputum AFB smear with patients with negative or unknown sputum AFB smear (e); of patients treated for PTB for the first time with patients with previous anti-TB treatment (f); and of patients with national ID number and patients without national ID number (g).
a
684 Table 3
S. de Valli` ere, R.D. Barker Mortality and hazard ratios for 295 patients admitted to hospital for pulmonary tuberculosis Mortality
Multivariate analysisa
Univariate analysis HR (95% CI)
P
HR (95% CI)
P-value
1.0 0.9 2.3 6.2 11.1
— 0.827 0.110 <0.001 <0.001
1.0 0.7 2.0 4.1 9.1
— 0.621 0.176 0.010 <0.001
Performance status 0 1 2 3 4
6/94 (6%) 3/55 (6%) 10/70 (14%) 9/27 (33%) 25/49 (51%)
Gender Male Female
29/177 (16%) 24/114 (21%)
1.0 1.3 (0.8—2.3)
— 0.318
— —
Age group (years) 20—34 35—49 ≥50
12/114 (11%) 26/119 (22%) 15/62 (24%)
1.0 2.3 (1.1—4.5) 2.4 (1.1—5.1)
— 0.019 0.024
1.0 2.3 (1.2—4.7) 2.0 (0.9—4.3)
— 0.017 0.070
Sputum AFB smear Positive Negative or unknown
30/229 (13%) 23/66 (35%)
1.0 3.0 (1.7—5.2)
— <0.001
1.0 2.3 (1.3—4.0)
— 0.004
Previous anti-TB treatment No Yes
46/252 (18%) 7/39 (18%)
1.0 1.0 (0.5-2.3)
— 0.936
— —
National ID number Unavailable Available
12/116 (10%) 41/179 (23%)
1.0 2.3 (1.2—4.4)
— 0.010
1.0 2.3 (1.2—4.4)
a
(0.2—3.4) (0.8—6.3) (2.2—17.5) (4.6—27.2)
(0.2—2.8) (0.7—5.6) (1.4—12.1) (3.7—22.3)
— —
— — — 0.013
For the multivariate analysis performance status, age group, sputum smear result and availability of national ID number were included.
respectively. The 2 test for linear trend analysing the mortality according to performance score was also highly significant (2 for linear trend = 47.4, P < 0.001). Univariate Cox regression analysis showed that the risk of death was also increased in older patients, in patients with negative or unknown sputum AFB smear status and in patients with a national ID number. However, there were no significant differences in mortality according to gender or history of previous treatment for TB. Multivariate Cox regression analysis showed that the performance status, sputum AFB smear results, and availability of a national ID number were independently associated with differences in mortality. The age category 35—49 years was also shown to be independently associated with a higher mortality. HIV status could not be used for the Cox regression analysis, as test results were only available for 32% of patients.
4. Discussion Death from TB represents a major burden to the societies of the poorer countries in the world. Ultimately, mortality from TB will be reduced by improving the regional and national TB control programmes, leading to earlier diagnosis and treatment of patients. This will probably be achieved through economical development that supports general health and health services infrastructure. Other problems such as access to TB treatment facilities, patient perceptions of disease and its care, stigma, level of awareness about TB and alternative care providers will also have
to be addressed. However, the current reality is that many hospitals, especially in sub-Saharan Africa, are overwhelmed with very sick TB patients, especially PTB patients with a high mortality rate, and this situation is unlikely to change rapidly. The 2-month mortality of 18% observed in this cohort of adult patients admitted for PTB to hospital and started on anti-TB treatment was comparable to other reports from South Africa (Alvarez et al., 2004; Edginton et al., 2005). It should be emphasised, however, that these figures did not represent the overall mortality from PTB in this region, because many patients with PTB were managed on an outpatient basis. The Kaplan Meier survival curves and the hazard ratios showed a strong association between poor performance status and risk of dying. Patients with a performance status of 3 or 4 had a particularly high risk of dying. Thirtythree percent of patients with a performance status of 3, and 51% of patients with a performance status of 4, died within 2 months of the start of anti-TB treatment. As can be seen from the Kaplan Meier survival curves, the great majority of patients died within the first month of starting anti-TB treatment. The fact that we saw a similarly decreased survival according to performance status in the sub-group of ss+ patients indicates that the observed relationship between performance status and death is not simply due to the inclusion of patients with unclear diagnosis in the poorer performance status categories. Besides poor performance status, negative or unknown sputum AFB smear was independently associated with a higher risk of dying. It is
Performance status in pulmonary tuberculosis probable that this was due to difficulties of obtaining good sputum from very sick patients, or due to a higher proportion of HIV-positive patients. The availability of a national ID number was also associated with poorer survival. It is likely that the national ID number was more frequently available for patients who died, because the recording of the national ID number is a legal requirement for completion of a death certificate. Therefore, the national ID number was possibly added to some data collection sheets after death. In this study, we focused on PTB patients and excluded cases of extrapulmonary TB for two reasons. Firstly, PTB cases represent by far the largest number of patients admitted to hospital for TB. Therefore, it appears particularly urgent to find innovative ways to decrease the death burden from this type of TB. Secondly, we wanted in a first step to evaluate the performance status on a relatively homogenous population. Nevertheless, we think that another study to evaluate the usefulness of the performance status in extrapulmonary TB might also be appropriate. It is unfortunate that HIV test results were available for only 32% of patients. For this reason, it was not possible to adjust for this variable in the statistical analysis. However, as an HIV test was not a requirement for study entry, it was not surprising that results were only available for such a small proportion of patients. In resource-limited countries, the prospect of a possibly positive HIV diagnosis is often a reason for refusing the test given the absence of affordable antiretroviral treatment. Among the patients tested, the seroprevalence was 79%, which is comparable to other reports from sub-Saharan Africa of HIV seropositivity in PTB patients (Edginton et al., 2005; Zachariah et al., 2002). Among the patients for whom an HIV test was done, all recorded deaths were seen in HIV-positive patients. It could therefore be argued that the performance score is just a surrogate marker for HIV positivity in this population. Even if this is the case, such a performance score would remain useful, as its use would assist in the immediate decision-making before the HIV test result is available. Another possible limitation of this study is that some deaths were missed and therefore the outcome of some cases was misclassified in the TB registers. Furthermore, the provincial administration may have been unable to identify patients who died because of variable spelling of names. These factors would probably have led to an underestimation of the number of deaths. However, we believe that it is unlikely that the relationship between performance status and mortality has been created by this possible misclassification bias. The ability to identify patients at high risk of death using a performance status could be a useful tool in both clinical and research situations. In the clinical setting, the systematic assessment of performance status might help to identify rapidly the sub-group of PTB patients that requires special attention, such as more vigorous resuscitation measures and more frequent medical assessment. It is our experience that PTB patients are frequently neglected in understaffed hospitals in countries with limited resources. Another important operational recommendation would be to improve diagnostic HIV counselling and testing for this group of patients. This may allow improved management of other opportunistic infections, providing intensive nutritional support and antiretroviral treatment.
685 Besides improved medical management of cases, there is also a need for renewed research into preventing death from TB, and there has been a specific call for research in this area (Harries et al., 2001). Several interventions aimed at reducing mortality could be tested using this performance status. Corticosteroids are considered lifesaving by some clinicians treating critically ill PTB patients, although this belief has never been scientifically confirmed (Crofton et al., 1999). It has also been suggested that PTB patients sometimes have dual infection with other pathogens (Schleicher and Feldman, 2003), and concomitant administration of an empiric broad-spectrum antibiotic could be evaluated. Increased doses of anti-TB drugs might be beneficial in HIV-positive patients because of problems of malabsorption (Stout et al., 2004). Selecting patients with a high mortality rate may improve our ability to demonstrate the effectiveness of such potentially lifesaving interventions. Several scoring systems are in use to assess severity and prognosis of different diseases. These scoring systems have mainly been developed to assess the risk of death in patients with malignancies and those admitted to intensive care units. We are not aware of any scoring system specifically designed to evaluate how sick a patient with TB is. We chose a simple scoring system to ensure that the doctors and nurses had no difficulty in determining the score after a brief assessment of the patients. This is important, because regions with a high burden of PTB are mostly countries with limited resources, and health care workers are often already overwhelmed by their clinical duties. As it takes only a couple of minutes to determine the performance status, this parameter can be obtained at essentially no cost and without increasing the workload unduly. In addition, the result is immediately available — unlike laboratory tests, which usually take time to be obtained. In conclusion, poor performance status showed a strong association with early death among adult patients admitted to hospital and treated for PTB. We think that the recording of performance status could be a useful clinical, epidemiological and research tool in the effort to reduce the mortality from PTB. It could assist in improving case management, as well testing new interventions. Conflicts of interest The authors have no conflicts of interest concerning the work reported in this paper.
Acknowledgements We thank Drs S. Cana, R. Gutierrez, S. Lopez, M. Matai, J.C. Rodriguez, A. Sanchez and J.P. Zaldivar, as well as the nursing staff from the participating hospitals, for their assistance in data collection. We also thank H. Israel for help with the statistical analysis.
References Alvarez, G.G., Thembela, B.L., Muller, F.J., Clinch, J., Singhal, N., Cameron, D.W., 2004. Tuberculosis at Edendale Hospital in Pietermaritzburg, KwaZulu Natal, South Africa. Int. J. Tuberc. Lung Dis. 8, 1472—1478.
686 Ashley, E.A., Johnson, M.A., Lipman, M.C., 2000. Human immunodeficiency virus and respiratory infection. Curr. Opin. Pulm. Med. 6, 240—245. Crofton, J., Horne, N., Miller, F., 1999. Clinical Tuberculosis, second ed. Macmillan Education, London. Dye, C., Scheele, S., Dolin, P., Pathania, V., Raviglione, M.C., for the WHO global surveillance and monitoring project, 1999. Global burden of tuberculosis: estimated incidence, prevalence, and mortality by country. JAMA 282, 677— 686. Edginton, M.E., Wong, M.L., Phofa, R., Mahlaba, D., Hodkinson, H.J., 2005. Tuberculosis at Chris Baragwanath Hospital: numbers of patients diagnosed and outcomes of referrals to district clinics. Int. J. Tuberc. Lung Dis. 9, 398—402. Harries, A.D., Hargreaves, N.J., Kemp, J., Jindani, A., Enarson, D.A., Maher, D., Salaniponi, F.M., 2001. Deaths from tuberculosis in sub-Saharan African countries with a high prevalence of HIV-1. Lancet 357, 1519—1523. Oken, M.M., Creech, R.H., Tormey, D.C., Horton, J., Davis, T.E., McFadden, E.T., Carbone, P.P., 1982. Toxicity and response
S. de Valli` ere, R.D. Barker criteria of the eastern cooperative oncology group. Am. J. Clin. Oncol. 5, 649—655. Schleicher, G.K., Feldman, C., 2003. Dual infection with Streptococcus pneumoniae and Mycobacterium tuberculosis in HIVseropositive patients with community acquired pneumonia. Int. J. Tuberc. Lung Dis. 7, 1207—1208. Stout, J.E., Weintrob, A.C., Engemann, J.J., Fortenberry, E.R., Peloquin, C.A., Hamilton, C.D., 2004. Therapeutic drug monitoring in patients with tuberculosis and advanced human immunodeficiency virus infection. Annual meeting of the Infectious Diseases Society of America, Boston, MA, 30 September—3 October 2004 (Abstract 905). WHO, 2005. Global Tuberculosis Control — Surveillance, Planning, Financing. World Health Organization, Geneva. http:// www.who.int/tb/publications/global report/en/ [accessed October 2005]. Zachariah, R., Spielmann, M.P., Harries, A.D., Salaniponi, M.L., 2002. Moderate to severe malnutrition in patients with tuberculosis is a risk factor associated with early death. Trans. R. Soc. Trop. Med. Hyg. 96, 291—294.
available at www.sciencedirect.com
journal homepage: www.elsevierhealth.com/journals/trst
Poor performance status is associated with early death in patients with pulmonary tuberculosis S. de Valli` ere a,∗, R.D. Barker b a b
Pietersburg-Mankweng Hospital Complex, Private Bag X9316, Polokwane 0700, South Africa Department of Respiratory Medicine, King’s College Hospital, Camberwell, London, UK
Received 6 May 2005 ; received in revised form 5 September 2005; accepted 5 September 2005 Available online 28 November 2005
KEYWORDS Pulmonary tuberculosis; Performance status; Early death; South Africa
Summary The objective of this study was to determine whether poor performance status at the start of anti-tuberculous (anti-TB) treatment is associated with early death in patients admitted to hospital with pulmonary tuberculosis (PTB). During 3 months in 2001, all adult patients admitted to eight hospitals in Limpopo Province, South Africa, and diagnosed with PTB were eligible for inclusion. At initiation of anti-TB treatment, a performance status between 0 and 4 was estimated for each patient using a modified version of the Eastern Cooperative Oncology Group scoring system. Hospital records and local TB registers were reviewed to identify patients who had died during the first 2 months of treatment. In addition, it was ascertained whether a death notification had been received by the provincial administration. Fifty-three of 295 (18%) patients died within 2 months. Mortality increased from 6% in patients with the best performance status to 51% in patients with the poorest performance status. Univariate and multivariate Cox regression analysis showed that the hazard ratio for dying was significantly higher for patients with a performance status of 3 or 4. Poor performance status shows a strong association with early death in patients with PTB and has the potential to be a useful clinical, epidemiological and research tool. © 2005 Royal Society of Tropical Medicine and Hygiene. Published by Elsevier Ltd. All rights reserved.
1. Introduction Despite the availability of potent chemotherapeutic agents, patients with tuberculosis (TB) continue to die. It is
∗ Corresponding author. Present address: Division of Infectious Diseases, St. Louis University Hospital, 3635 Vista Ave, FDT-8N, St. Louis, MO 63110, USA. Tel.: +1 314 577 8648; fax: +1 314 771 3816. E-mail addresses: [email protected] (S. de Valli` ere), [email protected] (R.D. Barker).
estimated that 2 to 3 million patients die from TB every year worldwide (Dye et al., 1999). According to WHO, the mortality of new sputum-smear-positive (ss+) patients should be less than 5%, but many countries do not achieve this goal (WHO, 2005). Of the 22 countries with the highest TB incidence in the world, 10 reported a mortality of new ss+ patients of over 5% (5.4—11%). Among those high-burden countries, South Africa reported a mortality of new ss+ cases of 7%. Co-infection with HIV has been identified as a major factor contributing to increased mortality in patients with pulmonary tuberculosis (PTB) (Ashley et al., 2000). However,
0035-9203/$ — see front matter © 2005 Royal Society of Tropical Medicine and Hygiene. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.trstmh.2005.09.007
682
S. de Valli` ere, R.D. Barker
many other factors, such as delay in diagnosis, inadequacy of chemotherapeutic regimens, concomitant diseases, malnutrition and others, might affect the outcome of treatment in patients with PTB. In 2001 the Global Fund to fight AIDS, TB and malaria was created with the aim of ensuring additional resources to reduce the morbidity and mortality from these three diseases. Reduction of death from TB is also an element of the ‘Millennium Development Goal’, which has a target to halve TB-associated mortality by 2015. It would be useful to have a simple tool to identify which patients with PTB are most at risk of dying. Such a tool could be used for clinical purposes, as well as to investigate the potential benefit of new interventions that have been proposed in the past (Harries et al., 2001). The objective of this study was to determine whether the performance status of PTB patients assessed at the time of initiation of anti-TB treatment was associated with death within 2 months.
2. Materials and methods The study was carried out in the Limpopo Province, South Africa, during 2001. There were 38 hospitals in the Limpopo Province, and eight were selected for the study based on their availability of medical and nursing staff to support the study and the knowledge that they were treating a significant number of patients with PTB. For 3 months, all adults aged 20 years or older admitted to those hospitals and put on anti-TB treatment for PTB were eligible for inclusion in the study. The diagnosis of PTB was considered established if the patient was entered into the local TB register as a PTB case. Before the start of the data collection, one of the authors visited all the participating hospitals and organized a training seminar with the staff that would be involved in the data collection. At the time anti-TB treatment was started, the doctor or the nurse in charge completed the first section of a data collection sheet. This section required the recording of the patient’s name, gender, date of birth, national ID number if available, date on which anti-TB treatment was started, history of previous treatment for TB and the estimation of a performance status. For this purpose, we used a modified version of the scoring system proposed by the Eastern
Cooperative Oncology Group (ECOG) (Table 1) (Oken et al., 1982). We added to each category a brief description of the type of activity the patients should be able to carry out, taking local conditions into consideration. In all participating hospitals, the TB wards were situated on the ground level. The bathrooms were in proximity to the patient rooms (less than 10 m), and the patients had direct access to a backyard. The authors of this paper were not involved with the management of the patients. On discharge from the hospital, or if the patient died, the second section of the data collection sheet was completed, which required the recording of the sputum acid-fast bacillus (AFB) smear results as positive, negative or unknown. Sputum smear was considered unknown if the patient was unable to produce a sputum specimen that could be sent for analysis to the laboratory. The HIV test result was also recorded if available. As this was an observational study, an HIV test result was not mandatory. Finally, date of death was recorded if applicable. Three months after completion of the initial data collection, the vital status within 2 months of starting anti-TB treatment of the enrolled patients was ascertained by crosschecking hospital records and local TB registers. As a supplement, it was ascertained whether the provincial administration had received a death notification. The provincial administration was supplied with the name, first name, date of birth and the national ID number, if available, of all the patients enrolled in the study. For statistical analysis, the software package EpiInfo version 3.3.2 (CDC, Atlanta, GA, USA) was used. The 2 test and the 2 test for linear trend were used to test for differences in proportions and linear trends, respectively. Kaplan Meier survival estimates were compared using the Cox-Mantel test (log-rank test). Univariate and multivariate Cox regression analyses were performed to calculate hazard ratios with 95% CI. For the multivariate analysis, a stepwise backward model was used. The performance status was the dependent variable. The variables with a significance level >0.05 were sequentially excluded.
3. Results During the 3-month study period, 343 patients were admitted for suspected PTB to the adult wards of the eight
Table 1 Definition of the five categories of the Eastern Cooperative Oncology Group (ECOG) performance scoring system and the additional descriptions used to take local conditions into consideration ECOG categories
Additional description of performance
Score
Able to carry out normal activity without restriction Unable to do physically strenuous activity, but ambulatory and able to carry out light work Ambulatory and capable of all self-care, but unable to carry out any work; up and about <50% of waking hours Capable of only limited self-care; confined to bed or chair >50% of waking hours Completely disabled; cannot carry out any self-care; totally confined to bed or chair
No physical restriction
0
Able to walk around the whole hospital premises, but unable to do any income-generating work Able to walk around the ward and to the backyard
1
Able to go to the bathroom, but unable to walk around the ward Unable to go to the bathroom
2
3 4
Performance status in pulmonary tuberculosis
683
participating hospitals. Forty-eight patients were excluded for the following reasons: no estimation of performance status (7 patients); age missing or less than 20 years (37 patients); and anti-TB treatment started before admission (4 patients). Characteristics of the 295 included patients are shown in Table 2: 61% were male; the median age was 37 (range 20—81) years; 229 (78%) patients were ss+; HIV test result was only available for 95 (32%) patients, but 75 (79%) of the tested patients were seropositive; 16 of 20 (80%) HIV-negative patients were ss+, while only 46 of 75 (61%) HIVpositive patients were ss+ — this difference was, however, not statistically significant (2 = 2.4, P = 0.119). On the performance status scale from 0 (best condition) to 4 (worst condition), 76 of 295 patients (26%) had a performance status of 3 or 4 (Table 2). Significantly poorer performance status was seen in women, HIV-positive subjects, and patients with negative or unknown sputum smears. Overall, 53 (18%) patients died within the study period. Forty-three (15%) patients died within 1 month, while 10 (3%) patients died during the second month of anti-TB treatment. The Kaplan Meier survival curves showed an increased risk of dying with poorer performance status (P < 0.001 by log-rank test) (Figure 1). A sub-analysis of the 229 sputum AFBpositive patients showed similar Kaplan Meier survival curves (data not shown). Looking at the entire cohort, mortality
Table 2
Figure 1 Kaplan Meier survival curves according to performance status for 295 patients admitted to hospital and started on anti-tuberculous treatment; P < 0.001 by the Cox-Mantel test (log-rank test). PS: performance status.
increased from 6% in the patients with the best performance status to 51% in the patients with the poorest performance status (Table 3). Univariate Cox regression analysis showed that patients with performance status of 3 and 4 had hazard ratios of 6.2 (95% CI 2.2—17.5) and 11.1 (95% CI 4.6—27.2),
Characteristics and performance status of 295 patients admitted to hospital and started on anti-tuberculous treatment n
P-valuesa
Performance status, n (%) 0
1
2
3
4
Gender Male Female Missing
177 114 4
71 (40) 23 (20)
27 (15) 27 (24)
40 (23) 28 (25)
14 (8) 12 (11)
25 (14) 24 (21)
Age group (years) 20—34 35—49 ≥50
114 119 62
40 (35) 36 (30) 18 (29)
23 (20) 21 (18) 11 (18)
29 (25) 28 (24) 13 (21)
7 (6) 9 (8) 11 (18)
15 (13) 25 (21) 9 (15)
0.263c
HIV Positive Negative Missing
75 20 200
11 (15) 11 (55) 72 (36)
10 (13) 3 (15) 42 (21)
17 (23) 6 (30) 47 (24)
13 (17) 0 (0) 14 (7)
24 (32) 0 (0) 25 (13)
<0.001d
Sputum AFB smear Positive Negative or unknown
229 66
80 (35) 14 (21)
43 (19) 12 (18)
59 (26) 11 (17)
17 (7) 10 (15)
30 (13) 19 (29)
0.003e
Previous anti-TB treatment No 252 Yes 39 Unknown 4
76 (30) 18 (46)
48 (19) 6 (15)
60 (24) 9 (23)
26 (10) 1 (3)
42 (17) 5 (13)
National ID number Unavailable Available
116 179
45 (39) 49 (27)
23 (20) 32 (18)
21 (18) 49 (27)
10 (9) 17 (9)
17 (15) 32 (18)
Total
295
94 (32)
55 (19)
70 (24)
27 (9)
49 (17)
0.008b
0.247f
0.194g
P-values of the 2 tests comparing the distribution of the performance status of: males with females (b); of the different age groups (c); of HIV-positive with HIV-negative patients (d); of patients with positive sputum AFB smear with patients with negative or unknown sputum AFB smear (e); of patients treated for PTB for the first time with patients with previous anti-TB treatment (f); and of patients with national ID number and patients without national ID number (g).
a
684 Table 3
S. de Valli` ere, R.D. Barker Mortality and hazard ratios for 295 patients admitted to hospital for pulmonary tuberculosis Mortality
Multivariate analysisa
Univariate analysis HR (95% CI)
P
HR (95% CI)
P-value
1.0 0.9 2.3 6.2 11.1
— 0.827 0.110 <0.001 <0.001
1.0 0.7 2.0 4.1 9.1
— 0.621 0.176 0.010 <0.001
Performance status 0 1 2 3 4
6/94 (6%) 3/55 (6%) 10/70 (14%) 9/27 (33%) 25/49 (51%)
Gender Male Female
29/177 (16%) 24/114 (21%)
1.0 1.3 (0.8—2.3)
— 0.318
— —
Age group (years) 20—34 35—49 ≥50
12/114 (11%) 26/119 (22%) 15/62 (24%)
1.0 2.3 (1.1—4.5) 2.4 (1.1—5.1)
— 0.019 0.024
1.0 2.3 (1.2—4.7) 2.0 (0.9—4.3)
— 0.017 0.070
Sputum AFB smear Positive Negative or unknown
30/229 (13%) 23/66 (35%)
1.0 3.0 (1.7—5.2)
— <0.001
1.0 2.3 (1.3—4.0)
— 0.004
Previous anti-TB treatment No Yes
46/252 (18%) 7/39 (18%)
1.0 1.0 (0.5-2.3)
— 0.936
— —
National ID number Unavailable Available
12/116 (10%) 41/179 (23%)
1.0 2.3 (1.2—4.4)
— 0.010
1.0 2.3 (1.2—4.4)
a
(0.2—3.4) (0.8—6.3) (2.2—17.5) (4.6—27.2)
(0.2—2.8) (0.7—5.6) (1.4—12.1) (3.7—22.3)
— —
— — — 0.013
For the multivariate analysis performance status, age group, sputum smear result and availability of national ID number were included.
respectively. The 2 test for linear trend analysing the mortality according to performance score was also highly significant (2 for linear trend = 47.4, P < 0.001). Univariate Cox regression analysis showed that the risk of death was also increased in older patients, in patients with negative or unknown sputum AFB smear status and in patients with a national ID number. However, there were no significant differences in mortality according to gender or history of previous treatment for TB. Multivariate Cox regression analysis showed that the performance status, sputum AFB smear results, and availability of a national ID number were independently associated with differences in mortality. The age category 35—49 years was also shown to be independently associated with a higher mortality. HIV status could not be used for the Cox regression analysis, as test results were only available for 32% of patients.
4. Discussion Death from TB represents a major burden to the societies of the poorer countries in the world. Ultimately, mortality from TB will be reduced by improving the regional and national TB control programmes, leading to earlier diagnosis and treatment of patients. This will probably be achieved through economical development that supports general health and health services infrastructure. Other problems such as access to TB treatment facilities, patient perceptions of disease and its care, stigma, level of awareness about TB and alternative care providers will also have
to be addressed. However, the current reality is that many hospitals, especially in sub-Saharan Africa, are overwhelmed with very sick TB patients, especially PTB patients with a high mortality rate, and this situation is unlikely to change rapidly. The 2-month mortality of 18% observed in this cohort of adult patients admitted for PTB to hospital and started on anti-TB treatment was comparable to other reports from South Africa (Alvarez et al., 2004; Edginton et al., 2005). It should be emphasised, however, that these figures did not represent the overall mortality from PTB in this region, because many patients with PTB were managed on an outpatient basis. The Kaplan Meier survival curves and the hazard ratios showed a strong association between poor performance status and risk of dying. Patients with a performance status of 3 or 4 had a particularly high risk of dying. Thirtythree percent of patients with a performance status of 3, and 51% of patients with a performance status of 4, died within 2 months of the start of anti-TB treatment. As can be seen from the Kaplan Meier survival curves, the great majority of patients died within the first month of starting anti-TB treatment. The fact that we saw a similarly decreased survival according to performance status in the sub-group of ss+ patients indicates that the observed relationship between performance status and death is not simply due to the inclusion of patients with unclear diagnosis in the poorer performance status categories. Besides poor performance status, negative or unknown sputum AFB smear was independently associated with a higher risk of dying. It is
Performance status in pulmonary tuberculosis probable that this was due to difficulties of obtaining good sputum from very sick patients, or due to a higher proportion of HIV-positive patients. The availability of a national ID number was also associated with poorer survival. It is likely that the national ID number was more frequently available for patients who died, because the recording of the national ID number is a legal requirement for completion of a death certificate. Therefore, the national ID number was possibly added to some data collection sheets after death. In this study, we focused on PTB patients and excluded cases of extrapulmonary TB for two reasons. Firstly, PTB cases represent by far the largest number of patients admitted to hospital for TB. Therefore, it appears particularly urgent to find innovative ways to decrease the death burden from this type of TB. Secondly, we wanted in a first step to evaluate the performance status on a relatively homogenous population. Nevertheless, we think that another study to evaluate the usefulness of the performance status in extrapulmonary TB might also be appropriate. It is unfortunate that HIV test results were available for only 32% of patients. For this reason, it was not possible to adjust for this variable in the statistical analysis. However, as an HIV test was not a requirement for study entry, it was not surprising that results were only available for such a small proportion of patients. In resource-limited countries, the prospect of a possibly positive HIV diagnosis is often a reason for refusing the test given the absence of affordable antiretroviral treatment. Among the patients tested, the seroprevalence was 79%, which is comparable to other reports from sub-Saharan Africa of HIV seropositivity in PTB patients (Edginton et al., 2005; Zachariah et al., 2002). Among the patients for whom an HIV test was done, all recorded deaths were seen in HIV-positive patients. It could therefore be argued that the performance score is just a surrogate marker for HIV positivity in this population. Even if this is the case, such a performance score would remain useful, as its use would assist in the immediate decision-making before the HIV test result is available. Another possible limitation of this study is that some deaths were missed and therefore the outcome of some cases was misclassified in the TB registers. Furthermore, the provincial administration may have been unable to identify patients who died because of variable spelling of names. These factors would probably have led to an underestimation of the number of deaths. However, we believe that it is unlikely that the relationship between performance status and mortality has been created by this possible misclassification bias. The ability to identify patients at high risk of death using a performance status could be a useful tool in both clinical and research situations. In the clinical setting, the systematic assessment of performance status might help to identify rapidly the sub-group of PTB patients that requires special attention, such as more vigorous resuscitation measures and more frequent medical assessment. It is our experience that PTB patients are frequently neglected in understaffed hospitals in countries with limited resources. Another important operational recommendation would be to improve diagnostic HIV counselling and testing for this group of patients. This may allow improved management of other opportunistic infections, providing intensive nutritional support and antiretroviral treatment.
685 Besides improved medical management of cases, there is also a need for renewed research into preventing death from TB, and there has been a specific call for research in this area (Harries et al., 2001). Several interventions aimed at reducing mortality could be tested using this performance status. Corticosteroids are considered lifesaving by some clinicians treating critically ill PTB patients, although this belief has never been scientifically confirmed (Crofton et al., 1999). It has also been suggested that PTB patients sometimes have dual infection with other pathogens (Schleicher and Feldman, 2003), and concomitant administration of an empiric broad-spectrum antibiotic could be evaluated. Increased doses of anti-TB drugs might be beneficial in HIV-positive patients because of problems of malabsorption (Stout et al., 2004). Selecting patients with a high mortality rate may improve our ability to demonstrate the effectiveness of such potentially lifesaving interventions. Several scoring systems are in use to assess severity and prognosis of different diseases. These scoring systems have mainly been developed to assess the risk of death in patients with malignancies and those admitted to intensive care units. We are not aware of any scoring system specifically designed to evaluate how sick a patient with TB is. We chose a simple scoring system to ensure that the doctors and nurses had no difficulty in determining the score after a brief assessment of the patients. This is important, because regions with a high burden of PTB are mostly countries with limited resources, and health care workers are often already overwhelmed by their clinical duties. As it takes only a couple of minutes to determine the performance status, this parameter can be obtained at essentially no cost and without increasing the workload unduly. In addition, the result is immediately available — unlike laboratory tests, which usually take time to be obtained. In conclusion, poor performance status showed a strong association with early death among adult patients admitted to hospital and treated for PTB. We think that the recording of performance status could be a useful clinical, epidemiological and research tool in the effort to reduce the mortality from PTB. It could assist in improving case management, as well testing new interventions. Conflicts of interest The authors have no conflicts of interest concerning the work reported in this paper.
Acknowledgements We thank Drs S. Cana, R. Gutierrez, S. Lopez, M. Matai, J.C. Rodriguez, A. Sanchez and J.P. Zaldivar, as well as the nursing staff from the participating hospitals, for their assistance in data collection. We also thank H. Israel for help with the statistical analysis.
References Alvarez, G.G., Thembela, B.L., Muller, F.J., Clinch, J., Singhal, N., Cameron, D.W., 2004. Tuberculosis at Edendale Hospital in Pietermaritzburg, KwaZulu Natal, South Africa. Int. J. Tuberc. Lung Dis. 8, 1472—1478.
686 Ashley, E.A., Johnson, M.A., Lipman, M.C., 2000. Human immunodeficiency virus and respiratory infection. Curr. Opin. Pulm. Med. 6, 240—245. Crofton, J., Horne, N., Miller, F., 1999. Clinical Tuberculosis, second ed. Macmillan Education, London. Dye, C., Scheele, S., Dolin, P., Pathania, V., Raviglione, M.C., for the WHO global surveillance and monitoring project, 1999. Global burden of tuberculosis: estimated incidence, prevalence, and mortality by country. JAMA 282, 677— 686. Edginton, M.E., Wong, M.L., Phofa, R., Mahlaba, D., Hodkinson, H.J., 2005. Tuberculosis at Chris Baragwanath Hospital: numbers of patients diagnosed and outcomes of referrals to district clinics. Int. J. Tuberc. Lung Dis. 9, 398—402. Harries, A.D., Hargreaves, N.J., Kemp, J., Jindani, A., Enarson, D.A., Maher, D., Salaniponi, F.M., 2001. Deaths from tuberculosis in sub-Saharan African countries with a high prevalence of HIV-1. Lancet 357, 1519—1523. Oken, M.M., Creech, R.H., Tormey, D.C., Horton, J., Davis, T.E., McFadden, E.T., Carbone, P.P., 1982. Toxicity and response
S. de Valli` ere, R.D. Barker criteria of the eastern cooperative oncology group. Am. J. Clin. Oncol. 5, 649—655. Schleicher, G.K., Feldman, C., 2003. Dual infection with Streptococcus pneumoniae and Mycobacterium tuberculosis in HIVseropositive patients with community acquired pneumonia. Int. J. Tuberc. Lung Dis. 7, 1207—1208. Stout, J.E., Weintrob, A.C., Engemann, J.J., Fortenberry, E.R., Peloquin, C.A., Hamilton, C.D., 2004. Therapeutic drug monitoring in patients with tuberculosis and advanced human immunodeficiency virus infection. Annual meeting of the Infectious Diseases Society of America, Boston, MA, 30 September—3 October 2004 (Abstract 905). WHO, 2005. Global Tuberculosis Control — Surveillance, Planning, Financing. World Health Organization, Geneva. http:// www.who.int/tb/publications/global report/en/ [accessed October 2005]. Zachariah, R., Spielmann, M.P., Harries, A.D., Salaniponi, M.L., 2002. Moderate to severe malnutrition in patients with tuberculosis is a risk factor associated with early death. Trans. R. Soc. Trop. Med. Hyg. 96, 291—294.