s
Tuberculosis case finding in a population with high HIV prevalence in western
Kenya
van 't Hoog, A.H.
Publication date
2012
Link to publication
Citation for published version (APA):
van 't Hoog, A. H. (2012). Tuberculosis case finding in a population with high HIV prevalence
in western Kenya.
General rights
It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulations
If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.
Excess Mortality and Risk Factors for Death in Adults
with Tuberculosis in Western Kenya
Anna H van’t Hoog1,2, J. Williamson1,3, Maquins Sewe1, Phelix Mboya4, M Lazarus O
Odeny1, Janet A Agaya1, Manase Amolloh1, Martien W Borgdorff2,5, Kayla F Laserson1,3
1 Kenya Medical Research Institute, KEMRI/CDC Research and Public Health Collaboration,
Kisumu, Kenya
2 Academic Medical Centre, University of Amsterdam, The Netherlands
3 Centers for Disease Control and Prevention, Center for Global Health, Atlanta, USA
4 Kenya Ministry of Health, Division of Leprosy, Tuberculosis and Lung Diseases, Kisumu,
Kenya
SuMMARy
Objectives
To estimate excess mortality among patients treated for TB in western Kenya between 2006 and 2008, relative to all-cause mortality in a population in the same geographic area with high HIV prevalence, and to evaluate risk factors for mortality during TB treatment. Methods
Clinical information was abstracted from surveillance registers of all patients of ≥15 years. We compared mortality rates during treatment with all-cause mortality from a health and demographic surveillance population in the same area to obtain standardized mortality ratios (SMR) and stratified mortality risk differences. Risk factors for excess mortality were obtained in a relative survival model and risk factors for death during treatment in a proportional hazards regression model.
Results
The crude mortality rate during TB treatment was 18.0 (95%CI 16.8-19.2) per 100 person years. Standardized for age and sex, the SMR was 8.8 (95%CI 8.2-9.4). Mortality was high in TB patients with unknown HIV-status (HR 2.9, 95%CI 2.2-3.8), or if HIV-infected not on antiretroviral therapy (ART) (HR 3.3; 95%CI 2.5-4.5), or not known to be on ART (HR 2.8; 95%CI 2.1-3.7). In relative survival analysis, excess mortality was greater in HIV-infected TB patients (excess hazard ratio 2.1, 95%CI 1.5-3.1), and lower in TB patients who were female or started treatment in a later year.
Conclusion
Increasing the uptake of existing TB-HIV interventions would enhance the decline in mortality during TB treatment. We estimated that among HIV-infected TB patients, 100% uptake would further reduce mortality by at least 31%.
4
InTRODuCTIOn
Death from tuberculosis (TB) is among the leading causes of premature death worldwide, especially among persons in the economically productive ages, in low and
middle income countries1. A 50% reduction in death rates by 2015 compared to the
1990 levels is one of the targets2 set to contribute to the Millennium Development Goals
to eradicate poverty3. In 2009, there were an estimated 1.7 million TB deaths, of which
0.38 million were in persons infected with HIV4, and mortality in TB patients had fallen by
around 35% since 1990. This reduction was not observed in the African region where HIV
contributes to high TB incidence and mortality5-7. TB-HIV interventions recommended
by the Stop TB Strategy since 2004 include HIV testing of TB patients, as an important prerequisite to providing HIV-infected TB patients with interventions to reduce mortality, including prophylaxis of opportunistic infections (cotrimoxazole [CPT]) and combination
antiretroviral therapy (ART)8, 9, but uptake is not yet optimal4.
In most populations with high burdens of TB and HIV, the estimation of TB mortality
has several limitations10. Weak or absent vital registration systems prohibit direct
measurement from causes of death10. Indirect mortality estimates rely on TB incidence,
which is uncertain, and on the case fatality ratio (CFR)10, 11. The CFR is defined as the
proportion of TB patients dying due to TB. However usually only the risk of death from
all causes during TB treatment is available12, which includes deaths from other causes.
An estimate of excess mortality provides adjustment for background mortality which will vary depending on the age structure of the population, HIV prevalence, and access to and quality of health care in the population. However, reliable stratified mortality rates from a comparable population are required but may be unavailable in resource limited settings.
In this study we evaluate excess TB mortality relative to all-cause mortality in a population
in the same geographic area in western Kenya, where TB and HIV are highly prevalent13-15.
Further, we evaluate risk factors for excess mortality. The comparison population is
monitored by a health and demographic surveillance system (HDSS)16, 17, which provides
details on population demography in areas where data from civil registration and health
service statistics are weak18. We also assess risk factors for death during TB treatment, to
METHODS
Study area
The study population is from a predominantly rural area in Nyanza Province, western Kenya. In 2007, HIV-prevalence among persons of 15-64 years old was 14.9% in the
province14, and TB notification was approximately 431/100,000 total population19,
considerably higher than in the rest of Kenya (338/100,000). Life expectancy at birth was
approximately 45 years17.
TB patients
Data were abstracted from the standard registers, on TB patients who were registered by the national TB control program (Division of Leprosy, Tuberculosis and Lung Diseases, DLTLD) in 2006, 2007 and 2008 at health facilities in the entire former Bondo and Siaya districts. In the clinic registers, nurses record information including age, sex, date treatment was started, type of TB and type of patient (new, relapse), sputum smear results, HIV status, and CPT was provided. At subsequent visits the registers are updated with the dates of the weekly visits (during intensive phase), the calendar months that the patient attended the clinic for drug collection (during continuation phase), and whether the patient is on treatment with combination antiretroviral therapy (ART). Final treatment outcomes are assigned by the district DLTLD coordinator. During the period 2006-2008 the duration of the standard treatment regimen for new adult TB patients was 8 months, although treatment was longer in case of an interruption. TB-HIV interventions including provider-initiated HIV testing for TB patients were introduced
from 200520, and HIV care and treatment services expanded around the same time21.
HDSS population
To determine excess mortality in TB patients we compared their mortality rates with stratified all-cause mortality rates of the population of the HDSS that is run by the Kenya Medical Research Institute and US Centers for Disease Control and Prevention. The HDSS includes the populations of three contiguous areas that cover approximately a quarter
of the former Bondo and Siaya Districts17. Asembo (former Bondo District) and Gem
(former Siaya District) have been included since 2001 and 2002 respectively16, and the
Karemo area (Siaya District) since 2007. From 2008, home-based HIV counseling and
testing (HBCT) was gradually introduced in the HDSS area22. The HIV status obtained
4
in/exclusion and follow up timeWe excluded TB patients aged less than 15 years or whose age was unknown, because in
young children the diagnosis of TB is less certain23, and the calculation of mortality rates
is more complex24. Patients with incomplete data on gender or type of TB, and patients
of whom ‘type of patient’ was unknown or were ‘transfer in’ were also excluded; the latter because of uncertainty about observed time on treatment and to avoid survival bias in this group.
For TB patients, the analysis time is the observed time on treatment that could be calculated from information available in the registers. Follow up time ended at the date of outcome, the last observed visit, or latest at 305 days (10 months) to allow for treatment interruptions. For records with follow-up time beyond 305 days, the treatment outcome was set at ‘still on treatment’, regardless of the outcome in the register. To manage missing data, the following adjustments were made (see Figure). When treatment start- or end-dates were unavailable or had obvious errors, they were estimated from the recorded visit dates during the intensive phase, and visit months during the continuation phase. When the records were insufficient to establish the duration of follow up (e.g. if only the treatment start date was reported), the following was applied: (i) if the treatment outcome was cure, completion or failure, the standard treatment duration of 8 months (243 days) was assumed; (ii) if the outcome was out of control, death or transfer out we assumed that the treatment initiation visit was observed, but the first weekly follow-up visit was not observed, and a duration of 4 days was imputed (middle of 1 week); (iii) if the outcome was missing, the record was excluded.
Annual all-cause mortality rates were obtained of HDSS residents of 15 years and older (n=175,037) for 2006, 2007 and 2008. To estimate the effect of HIV on excess mortality in TB patients, we used data on 33,926 (19%) HDSS residents who participated in HBCT in the course of 2008 and 2009. Of those 4,944 (14.6%) were HIV-infected.
Statistical analysis
Mortality rates per 100 person years (py) and corresponding 95% confidence intervals (CI) were obtained using Poisson regression. We calculated standardized mortality ratios (SMR) through direct standardization for sex and age-categories, calendar year, and HIV status, and calculated stratified mortality rate differences between TB patients and the HDSS population.
We used a relative survival model25 to estimate excess hazard ratios (eHR), which are
similar to Cox hazard ratios, but adjusted for expected background mortality. The advantage is that information on cause of death is not required, nor whether the excess
mortality is directly or indirectly attributable to the condition25, 26. The number of deaths
among TB patients in each demographic stratum was modeled in a generalized linear model with a Poisson error structure based on collapsed data, offsetting the expected deaths in similar strata of the HDSS population. The probability of death for each HDSS stratum of sex, age and calendar year was calculated as the number of deaths divided by the midyear population in each 1-year age group. The excess hazards are assumed to
be constant within pre-specified time periods25, which we took as one year. To assess the
eHR associated with HIV, we were restricted to data on HDSS residents with known HIV status from HBCT. Time with known HIV status was available for 22,487 py of follow up, providing on average the equivalent of 0.7 py per resident, which is similar to the average available per resident for one year for the whole HDSS population. The probability of death, stratified by age, sex and HIV were computed as described above, but taking the
population on 1st January 2009 as the midyear population.
Risk factors for death during Tb treatment
Risk factors for faster time to death during TB treatment were examined in univariable and multivariable Cox proportional hazards models. Patients who did not die were censored as described under ‘follow up time’. Interactions with HIV status and gender were examined. Variables were entered in the multivariable model if the univariable p-value was <0.20. On visual inspection of the log-minus log survival plots there was no evidence that the assumption of proportional hazards was violated for any of the co-variates. SAS 9.2 was used for statistical analysis.
RESulTS
During 2006-2008, the health facilities in Bondo and Siaya districts registered 10,876 patients aged 15 years or older for TB treatment (Figure 1). Of those, 1,495 (14%) patients were excluded because they were either a ‘transfer in’, or had missing values. Included and excluded patients had similar distributions of age, sex, type of patient, district, and treatment outcome. The excluded patients had more missing values for HIV and were slightly more often from the 2008 cohort (Table 1).
Among the 9,381 TB patients included in the analysis, 840 (9.0%) died, of whom 466 (55.5%) died within the first 2 months of treatment. The crude mortality rate during
4
treatment was 18.0 per 100 py (95%CI 16.8-19.2) (Table 2). Standardized for age andsex, the SMR was 8.8 (95%CI 8.2-9.4). The mortality rate ratios of TB patients relative to the HDSS population were highest in 15-24 old patients and decreased with increasing age. The rate difference increased with increasing age, and was highest (36.3/100 py; 95%CI 30.8-41.9) in male patients 65 years and older. Restricted to persons with known HIV status only, the HIV-standardized mortality ratio was 4.3 (95%CI 4.0-4.7), and the risk difference in HIV-uninfected TB patients was 6.3 per 100 py (95%CI 5.5-7.2).
Figure 1. Selection of TB patients
Age -below 15 years n=1,135 (9.3%) -missing n=145 (1.2%) Type of Patient -‘transfer in’ n=820 (8.3%) or -missing n=197 (1.8%) TB patients in analysis n=9,381 (86.3%) Missing values for
-Gender n=1 (0.01%) -Type of TB n=164 (1.5%)
Observed follow up 0 days (≤1 date recorded) or could not be calculated§
n=657 (6.0%) Treatment outcome ‘missing’,
-and only 1 visit date available (n=291; 2.7%)a,
-or all dates missing (n=22; 0.2%)
Calculated observed follow up time ≥ 1 day, regardless of treatment outcome§
n=9,037(83.1%)
Treatment outcome: ‘completed’, ‘cure’ or ‘failure’ → follow-up time set at 243 days (n=68; 0.6%)b
Treatment outcome: ‘out of control’, ‘transfer out’, or ‘death’: → follow-up time set at 4 days (n=276; 2.5%)c
TB patients with age 15 years and older n=10,876*
TB=Tuberculosis
*Denominator for all percentages below this box
§Follow-up time was the time between the treatment start-date and the date of outcome. If any of those dates were missing or had obvious entry or recording errors, adaptations were made as described in the methods section. A treatment start date was obtained for >99% of records. A treatment end date was obtained from the ‘date of outcome’ 64% of the records; for 26% from the last continuation phase visit moth, and for 9% a last visit date during the intensive phase. If the resulting follow-up time was zero, or irresolvable records were excluded or values imputed as described in the figure.
All TB patients registered at health facilities in the former Bondo and Siaya districts with year of registration 2006, 2007 or 2008: n=12,156
Table 1. Comparison of TB patients (age 15 years or older) in- and excluded in the analysis
Included Excluded Total p-value
Characteristic N row % N row % N
All patients 9,381 86% 1,495 14% 10,876 Gender Female 4,987 86% 824 14% 5,811 0.15* Male 4,394 87% 670 13% 5,064 missing* 0 0 1 100% 1 Age (years) 15-24 1,853 86% 299 14% 2,152 0.21 25-34 3,323 85% 574 15% 3,897 35-44 2,060 87% 317 13% 2,377 45-54 1,098 88% 156 12% 1,254 55-64 592 87% 86 13% 678 65+ 455 88% 63 12% 518 HIV-status Negative 1,597 88% 210 12% 1,807 <0.001 Positive 5,258 87% 765 13% 6,023 Unknown 2,526 83% 520 17% 3,046
Antiretroviral therapy in HIV-infected
HIV-positive ART-yes 1,171 86% 190 14% 1,361 0.06 HIV-positive ART-no 1,231 89% 152 11% 1,383 HIV-positive ART-unknown 2,856 87% 423 13% 3,279 Type of TB
PTB Smear-positive 3,301 89% 400 11% 3,701 0.03* PTB Smear-negative or not done 4,198 87% 610 13% 4,808
Extra-pulmonary TB 1,882 88% 261 12% 2,143
missing* 0 0% 224 100% 224
Type of Patient
New 8,460 95% 432 5% 8,892 0.89*
Retreatment/Relapse 921 95% 46 5% 967
4
Included Excluded Total p-value
Characteristic N row % N row % N
Year of Start Treatment
2006 3,129 87% 478 13% 3,607 <0.001 2007 3,196 88% 444 12% 3,640 2008 3,056 84% 573 16% 3,629 District Bondo 3,931 86% 659 14% 4,590 0.11 Siaya 5,450 87% 836 13% 6,286 Treatment outcome*
Treatment completed / Cure 5,585 90% 645 10% 6,230 0.36*
Default 1,051 88% 143 12% 1,194 Death 849 90% 98 10% 947 Failure 25 96% 1 4% 26 Transfer out 482 91% 50 9% 532 Not indicated* 1,389 71% 558 29% 1,947 PTB=pulmonary tuberculosis
Table 2. Mort ality r at es in TB pa tien ts, s tandar diz ed mort ality r atio’ s and mort ality r at e diff er ences TB pa tien ts HDSS mort ality per 100 py Expect ed dea ths Ra te diff er ence per 100 p y / 95% CI* Char act er -istic N % Ob ser ved dea ths py Mort ality r at e per 100 p y (95% CI) (S)MR 95% CI* All pa tien ts 9381 100% 840 4679 18.0 (16.8 ;19.2) 1.87 87.5 9.6 (9.0 ;10.3) 16.1 (15.7 ;16.5) Females, Ag e (y ear s) 15-24 1255 13% 76 617 12.3 (9.8; 15.4) 0.59 3.7 20.8 (16.4 ;26.0) 11.7 (11.1 ;12.4) 25-34 1738 19% 142 874 16.3 (13.8; 19.2) 1.84 16.1 8.8 (7.5 ;10.4) 14.4 (13.4 :15.4) 35-44 943 10% 88 495 17.8 (14.4; 21.9) 1.67 8.3 10.7 (8.5 ;13.1) 16.1 (14.9 ;17.3) 45-54 529 6% 39 276 14.1 (10.3; 19.4) 1.60 4.4 8.9 (6.3 ;12.1) 12.5 (11.0 ;14.1) 55-64 299 3% 15 156 9.6 (5.8; 16.0) 2.09 3.3 4.6 (2.6 ;7.6) 7.5 (5.2 ;9.8) 65+ 223 2% 26 108 24.2 (16.5; 35.5) 5.99 6.4 4.0 (3.0 ;6.5) 18.2 (13.5 ;22.9) Males, Ag e (y ear s) 15-24 598 6% 30 291 10.3 (7.2; 14.7) 0.34 1.0 30.6 (20.7 ;43.7) 10.0 (9.3 ;10.7) 25-34 1585 17% 142 779 18.2 (15.5; 21.5) 2.02 15.7 9.0 (7.6 ;10.7) 16.2 (15.1 ;17.3) 35-44 1117 12% 132 544 24.3 (20.5; 28.8) 2.82 15.3 8.6 (7.2 ;10.2) 21.4 (19.9 ;23.0) 45-54 569 6% 75 284 26.4 (21.1; 33.1) 3.06 8.7 8.6 (6.8 ;10.8) 23.4 (21.2 ;25.5) 55-64 293 3% 30 154 19.5 (13.6; 27.9) 3.14 4.8 6.2 (4.2 ;8.9) 16.4 (13.5 ;19.3) 65+ 232 2% 45 102 44.2 (33.0; 59.2) 7.88 8.0 5.6 (4.1 ;7.5) 36.3 (30.8 ;41.9) Ag e-Se x s tandar diz ed 840 95.6 8.8 (8.2 ;9.4)
4
TB pa tien ts HDSS mort ality per 100 py Expect ed dea ths Ra te diff er ence per 100 p y / 95% CI* Char act er -istic N % Ob ser ved dea ths py Mort ality r at e per 100 p y (95% CI) (S)MR 95% CI* year of St art T rea tmen t 2006 3129 33% 359 1607 22.3 (20.1; 24.8) 2.26 36.3 9.9 (8.9 ;11.0) 20.1 (19.3 ;20.9) 2007 3196 34% 272 1616 16.8 (14.9; 19.0) 1.80 29.0 9.4 (8.3 ;10.6) 15.0 (14.3 ;15.7) 2008 3056 33% 209 1456 14.4 (12.5; 16.4) 1.85 27.0 7.7 (6.7 ;8.9) 12.5 (11.8 ;13.2) Year St art T rea tmen t st andar diz ed 840 92.4 9.1 (8.5 ;9.7) HIV -s ta tus Neg ativ e 1597 17% 65 859 7.6 (5.9; 9.7) 1.24 † 10.6 6.1 (4.7 ;7.8) 6.3 (5.5 ;7.2) Positiv e 5258 56% 517 2,681 19.3 (17.7; 21.0) 4.67 † 125.2 4.1 (3.8 ;4.5) 14.6 (12.9 ;16.3) Unkno wn 2526 27% 258 1,139 22.6 (20.0; 25.6) HIV -s tandar diz ed (e xcluding ‘Unkno wn’) 582 135.8 4.3 (4.0 ;4.7) TB=T uber culosis; CFR=Case Fa tality Ra te, de fined as dea ths fr om an y cause during the ob ser ved tr ea tmen t period; py=per son year s; HDSS=Health and Demogr aphic Sur veillance S ys tem; SMR=s tandar diz ed mort ality r atio * 42. †Based on 22,487 py of follo w up with kno wn HIV results among 33,9 26 (19%) of HDSS residen ts of 15 year s and older whose HIV -s ta tus w as kno wn from Home based HIV c
ounseling and t
es
ting in 2008 and 2009. Of those 4,944 (14.6%) w
er
e HIV
-in
fect
HIV infection was a risk factor for excess mortality when adjusted for sex and age (eHR 2.1 (95% CI 1.5-3.1) (Table 3). Female TB patients experienced less excess mortality compared to males (eHR 0.78, 95%CI 0.59-1.0 when adjusted for HIV; eHR 0.75, 95%CI 0.52-1.1 when adjusted for age and calendar year). Age was not associated with excess mortality when adjusted for the other variables in the models. Between 2006 and 2008, mortality declined in the general HDSS population and in TB patients, both in TB patients who were HIV-infected, HIV-uninfected, and patients whose HIV status was unknown (not shown). Excess mortality was significantly lower in patients starting TB treatment in 2007 (eHR 0.65, 95%CI 0.43-0.96) and 2008 (eHR 0.40, 95%CI 0.24-0.69), compared to 2006.
Table 3. Risk factors for excess mortality in TB patients A*. Model with sex, age and year of
treat-ment start (unadjusted for HIV) B
†. Model with sex, age and HIV, all TB
pa-tients with known HIV-status of 2006-2008 combined
Relative Excess Risk (95% CI) Relative excess Risk (95% Ci)
Sex Gender
Female 0.75 (0.52 ; 1.1) Female 0.78 (0.59 ;1.0)
Male 1 Male 1
Age (years) Age (years)
15-24 1 15-24 1 25-34 0.80 (0.51 ; 1.3) 25-34 1.0 (0.70 ;1.5) 35-44 1.0 ( 0.61 ; 1.6) 35-44 1.1 (0.77 ;1.7) 45-54 0.86 ( 0.46 ; 1.6) 45-54 0.82 (0.47 ;1.4) 55-64 0.38 ( 0.10 ; 1.4) 55-64 0.59 (0.26 ;1.4) 65+ 0.51 (0.09 ; 3.1) 65+ 1.5 (0.66 ;3.2)
Year of Start Treatment HIV-status
2006 1 Negative 1
2007 0.65 (0.43 ; 0.96) Positive 2.1 (1.5 ;3.1)
2008 0.40 (0.24 ; 0.69)
CI=Confidence Interval; *HDSS population data of 2006-2008 used to offset population mortality †.HDSS data with known HIV status in 2008/2009 used to offset population mortality During TB treatment, female patients had lower mortality (Table 4). Stratified by HIV-status, this effect was only significant in HIV-infected patients (adjusted hazard ratio (aHR) 0.64; 95% CI 0.54-0.76). Although there were some interactions between sex and other risk factors, the aHR’s showed the same direction and combined aHRs are
4
presented in Table 4. Eighty-five percent of all deaths in 2006, and 77% in 2008, wereamong patients whose HIV status was unknown, or if HIV-infected, were not on ART, or not known to be on ART. In male patients, the aHR’s were 3.1 (95%CI 2.1-4.6) for HIV status unknown, and 4.5 (95%CI 3.0-6.8), and 3.5 (95% CI 2.4-5.1), respectively for not on, or not known to be on ART, compared to HIV-uninfected male patients. Among HIV-infected female patients on ART, the aHR was 1.6 (95% CI 0.99-2.5) compared to HIV-uninfected female patients. Among HIV-uninfected TB patients, mortality increased with age (not shown), and did not differ by gender (aHR 1.0, 95%CI 0.62-1.7). Strong risk factors for death were retreatment (aHR 3.0, 95% CI 1.6-5.7) and extra-pulmonary TB (aHR 5.1, 95%CI 2.4-10.9).
Of 6,855 TB patients with a known HIV status, 5,258 (79%) were HIV-infected of whom 4469 (85%) had documentation of CPT receipt or not; of those, 4302 (96%) had actually received CPT. The proportion of TB patients with unknown HIV-status decreased from 33% in 2006 to 24% 2008, and the proportion of HIV-infected patients with a record of being on ART increased from 18% to 26%. Both trends were similar in males and females (data not shown). Assuming similar HIV prevalence in patients with unknown
HIV status27, approximately 20% of HIV-infected TB patients were recorded to be on ART
in 2008.
Among TB patients with unknown and positive HIV status combined, the attributable fraction of incomplete uptake of HIV testing and ART to mortality was 64% (95%CI 54%-72%) when compared to the mortality rate in HIV-uninfected, and 31% (95%CI 15%-45%) when compared to HIV-infected patients on ART (Table 5).
DISCuSSIOn
This study showed that among TB patients on treatment in western Kenya, the risk of death was 8.8 times increased compared to the general population when standardized for sex and age, and 4.3 times when standardized for HIV. Adjusted for sex and age, HIV-infected TB patients were at increased risk of death compared to HIV-HIV-infected persons in the general population, while excess mortality was lower in TB patients who were female or started treatment at a later year. Among TB patients, unknown HIV-status, and if HIV-infected, not on ART, or not known to be on ART, were strong risk factors for death. Of the deaths among patients with a positive or unknown HIV status, approximately 31% could be reduced by improving uptake of HIV testing and ART, This fraction would be even higher if ART were initiated earlier.
Table 4. Risk f act or s f or dea th in pa tien ts on TB tr ea tmen t Follo w -up (p y) Mort ality r at e per 100p y (95% CI) Haz ar d Ra tio Patients Dea ths U ni va ria bl e Adjus ted (95% CI) All pa tien ts 9381 840 4,679 18.0 (16.8;19.2) Gender Female 4987 386 2,525 15.3 (13.8;16.9) 0.73 0.74 (0.65;0.85) Male 4394 454 2,154 21.1 (19.2;23.1) 1 1 Ag e (y ear s) 15-24 1747 106 908 11.7 (9.6;14.1) 1 1 25-34 3039 284 1,653 17.2 (15.3;19.3) 1.5 1.2 (0.99;1.5) 35-44 1840 220 1,039 21.2 (18.5;24.2) 1.8 1.5 (1.2;1.9) 45-54 984 114 559 20.4 (17.0;24.5) 1.8 1.6 (1.2;2.0) 55-64 547 45 310 14.5 (10.8;19.5) 1.3 1.2 (0.86;1.7) 65+ 384 71 209 33.9 (26.9;42.8) 2.9 3.1 (2.3;4.2) HIV -AR T HIV -neg ativ e 1597 65 859 7.6 (5.9;9.7) 1 1 HIV -positiv e AR T-yes 1171 94 652 14.4 (11.8;17.6) 1.9 2.0 (1.5;2.8) HIV -positiv e AR T-no 1231 140 612 22.9 (19.4;27.0) 2.9 3.3 (2.4;4.5) HIV -positiv e AR T-unkno wn 2856 283 1,417 20.0 (17.8;22.4) 2.6 2.8 (2.1;3.6) HIV -s ta tus unkno wn 2526 258 1,139 22.6 (20.0;25.6) 2.8 2.8 (2.1;3.7) Type of TB PTB Smear -positiv e 3301 238 1,699 14.0 (12.3;15.9) 1 1 PTB Smear -neg ativ e or not done 4198 417 2,070 20.1 (18.3;22.2) 1.4 1.5 (1.2;1.7) Ex tr apulmonar y TB 1882 185 910 20.3 (17.6;23.5) 1.4 1.5 (1.3;1.9)
4
Follo w -up (p y) Mort ality r at e per 100p y (95% CI) Haz ar d Ra tio Patients Dea ths U ni va ria bl e Adjus ted (95% CI) Type of P atien t N ew 8460 722 4,235 17.1 (15.9;18.3) 1 1 Re tr ea tmen t 921 118 444 26.6 (22.2;31.8) 1.5 1.6 (1.3;1.9) Year of T rea tmen t St art 2006 3129 359 1,607 22.3 (20.1;24.8) 1 1 2007 3196 272 1,616 16.8 (14.9;19.0) 0.8 0.7 (0.6;0.9) 2008 3056 209 1,456 14.4 (12.5;16.4) 0.6 0.6 (0.5;0.7) Dis trict Bondo 3931 327 1,907 17.1 (15.4;19.1) 1 Sia ya 5450 513 2,771 18.5 (17.0;20.2) 1.1 TB=T uber culosis; CFR=Case Fa tality Ra te, de fined as dea ths fr om an y cause during the ob ser ved tr ea tmen t period; p y=per son year s; HR= Hazard Ratio; AR T=AntiretroviralTreatment (combination therapy); CI=Confidence Interval; PTB=pulm
onary tuberculosis. Table 5. A ttribut able fr actions due t o inc omple te up tak e of HIV -tes ting and AR T among TB pa tien ts HIV and AR T s ta tus Pa tien ts (N, % of all pa tien ts) Dea ths (N, % of all dea ths) aHR Attribut able fr action (95% CI) aHR Attribut able fr action (95% CI) HIV neg ativ e 1597 17% 65 8% 1 baseline HIV -positiv e AR T-yes 1171 12% 94 11% 2.0 51% 32% 64% 1 baseline HIV -positiv e AR T-no 1231 13% 140 17% 3.3 70% 59% 78% 1.6 38% 20% 52% HIV -positiv e AR T-unkno wn 2856 30% 283 34% 2.8 64% 52% 72% 1.4 26% 6% 42% HIV unkno wn 2526 27% 258 31% 2.8 64% 53% 73% 1.4 28% 8% 43% HIV -positiv
e and HIV unkno
wn c ombined 7784 83% 775 92% 2.8 64% 54% 72% 1.5 31% 15% 45% AR T=Antiretroviral
Treatment (combination therapy);
TB=T
uberculosis; aHR=
Between 2006 and 2008, mortality and excess mortality in TB patients declined, likely explained by the rapid introduction and scale up of provider-initiated HIV testing, CPT and ART for TB patients in Kenya, and the expansion of HIV care and treatment services
in the region21. The decline is consistent with declining TB/HIV mortality rates in other
African countries with a generalized HIV-epidemic28.
Mortality in HIV-uninfected TB patients is still appreciable, although lower compared to
some other areas7, and the decline may reflect nationwide efforts to improve TB care.
A reduction of TB incidence in the population would further prevent these deaths. As in other studies, mortality in older TB patients was very high, and in part due to
co-morbidities7, 29.
The study findings reinforce the global target that 100% of HIV-infected TB patients
should be on ART by 2015 4 and the increasing call30 to initiate ART early in TB treatment,
and at higher CD4 cell counts31, 32 The low proportion of TB patients recorded to be
on ART is also explained by prioritization of individuals with more advanced immune
suppression9, and the delay of ART initiation until TB treatment, or at least the 2-month
intensive phase has been completed33, 34 to circumvent the complexities of concurrent
administration of TB and HIV treatment35. Furthermore, since TB-HIV services are not
fully integrated and HIV treatment services are not as decentralized as TB treatment
services, patients’ access to ART may be compromised36, 37.
Although increased mortality in male TB patients has been found in other studies7, 29,
in this study mortality did not differ by sex in TB patients with a negative or unknown HIV-status. In the population, uptake of HIV testing and treatment has been greater by
women22. Although referrals from HIV-clinics did not differ by gender (not shown), more
female TB patients may already have been on ART prior to TB diagnosis, contributing to lower excess mortality in females, and low mortality in female TB patients on ART. Early HIV case finding and initiation of ART would in addition to preventing HIV- and
TB-transmission30, further reduce TB mortality in HIV-infected persons who develop TB6.
This study has several limitations. Routine health service records suffer from missing data. This would mostly result in an underestimation of mortality in TB patients, and of the risk associated with lack of uptake of HIV testing and ART, for the following reasons: Mortality in patients who were lost to follow up or transferred was assumed to be similar
4
under reported due to the physical separation of TB and HIV clinics. Clinically diagnosedsmear-negative TB may include misclassifications39, 40, and restricted to smear-positive
TB patients, the risk of death from unknown or positive HIV status was stronger. The inclusion of TB deaths in all-cause mortality rates underestimates excess mortality, but was estimated to be less than 5%.
The effect of HIV on excess mortality should be interpreted with some caution. Since HIV-infected TB patients have been a priority for ART, better access to ART in TB patients would be expected compared to HIV-infected individuals in the general population, resulting in a low estimated effect of HIV. However, the effect could be an overestimate if, compared to TB patients, the HDSS population with known HIV status had less advanced immunosuppression, or if a larger proportion was on ART. In 2008 in general, access to ART was likely somewhat better compared to 2006. However, when restricted to TB patients from 2008 only, the HIV standardized mortality ratio was only slightly lower (3.6; 95% CI 3.0-4.2), and the excess HR reduced by less than 10%. Moreover, in the HDSS, less than half of HIV-infected individuals sought care within 6 months after
HBCT22. The effect of HIV may in part reflect lower ART utilization due to the delay in
starting ART in the presence of TB.
Incorporating prospective linkage of demographic, TB and HIV clinical data in the HDSS infrastructure, would allow better monitoring of the impact of TB and HIV interventions on mortality at a population level. At programmatic level, with increasing use of electronic
TB registers41, the distributions of TB patients and deaths by HIV and ART status could be
useful indicators for monitoring of TB-HIV deaths.
In conclusion, of the risk factors for death among TB patient in western Kenya, the effects of age, sex and decline in mortality over time are –at least in part- due to trends in the general population. Improving uptake of HIV testing of TB patients and ART to 100% could further reduce deaths by at least 31% among TB patients with a positive or unknown HIV status, and more if ART were also initiated earlier.
ACKnOWlEDGEMEnTS
We acknowledge the staff at the TB clinics and district TB coordinators, the KEMRI/CDC Research and Public Health Collaboration staff in general and the TB branch and HDSS branch in particular. This paper is published with the approval of the acting Director KEMRI. KEMRI/CDC HDSS is a member of the INDEPTH Network.
Disclaimer
The views expressed in this manuscript do not necessarily reflect the official policies of the Department of Health and Human Services; nor does mention of trade names, commercial practices or organizations imply endorsement by the US government. Sources of support
This work received support from the KEMRI/CDC Research and Public Health Collaboration and the Academic Medical Center of the University of Amsterdam
Author contributions
Study conception and design: AHH, MWB, KFL Study planning and conduct: AHH, JAA, LOO, PM, MA Data analysis: AHH, JW, SM
Drafting of manuscript: AHH
4
REFEREnCES
1. Lopez AD, Mathers CD, Ezzati M, Jamison DT, Murray CJ. Global Burden of Disease and Risk Factors. Washington (DC): World Bank; 2006.
2. Dye C, Maher D, Weil D, Espinal M, Raviglione M. Targets for global tuberculosis control. Int J Tuberc Lung Dis. 2006; 10(4): 460-2.
3. United Nations. Millennium Development Goals. 2000 [cited 2011 June 4th 2011]; Available from: http://www.un.org/ millenniumgoals/
4. World Health Organization. Global tuberculosis control: WHO report 2010. Geneva: World Health Organization (WHO/HTM/TB/2010.7); 2010.
5. Mukadi YD, Maher D, Harries A. Tuberculosis case fatality rates in high HIV prevalence populations in sub-Saharan Africa. Aids. 2001; 15(2): 143-52.
6. Straetemans M, Bierrenbach AL, Nagelkerke N, Glaziou P, van der Werf MJ. The effect of tuberculosis on mortality in HIV positive people: a meta-analysis. PLoS. 2010; 5(12): e15241.
7. Waitt CJ, Squire SB. A systematic review of risk factors for death in adults during and after tuberculosis treatment. Int J Tuberc Lung Dis. 2011; 15(7): 871-85. Epub 2011 Apr 12.
8. World Health Organization Stop TB partnership. The Stop TB Strategy; Building on and enhancing DOTS to meet the TB-related Millenium Development Goals. Geneva: World Health Organization (WHO/HTM/TB/2006.368); 2006.
9. World Health Organization STDaDoHA. Interim policy on collaborative TB/HIV activities. Geneva, Switzerland: World Health Organization, Stop TB Department and Department of HIV/AIDS; 2004..
10. Korenromp EL, Bierrenbach AL, Williams BG, Dye C. The measurement and estimation of tuberculosis mortality. Int J Tuberc Lung Dis. 2009; 13(3): 283-303. 11. Dye C, Bassili A, Bierrenbach AL,
Broekmans JF, Chadha VK, Glaziou P, et al. Measuring tuberculosis burden, trends, and the impact of control programmes. Lancet Infect Dis. 2008; 8(4): 233-43. 12. Straetemans M, Glaziou P, Bierrenbach AL,
Sismanidis C, van der Werf MJ. Assessing tuberculosis case fatality ratio: a meta-analysis. PLoS. 2011; 6(6): e20755. . 13. van’t Hoog AH, Laserson KF, Githui WA,
Meme HK, Agaya JA, Odeny LO, et al. High Prevalence of Pulmonary Tuberculosis and Inadequate Case Finding in Rural Western Kenya Am J Respir Crit Care Med. 2011; 183: 1245-53.
14. National AIDS/STI Control Programme (NASCOP) Kenya. 2007 Kenya AIDS Indicator Survey: Final Report. Nairobi: NASCOP; September 2009.
15. Amornkul PN, Vandenhoudt H, Nasokho P, Odhiambo F, Mwaengo D, Hightower A, et al. HIV prevalence and associated risk factors among individuals aged 13-34 years in Rural Western Kenya. PLoS One. 2009; 4(7): e6470.
16. Adazu K, Lindblade KA, Rosen DH, Odhiambo F, Ofware P, Kwach J, et al. Health and demographic surveillance in rural western Kenya: a platform for evaluating interventions to reduce morbidity and mortality from infectious diseases. Am J Trop Med Hyg. 2005; 73(6): 1151-8.
17. KEMRI/CDC Research and Public Health Collaboration. KEMRI/CDC Health and Demographic Surveillance System (HDSS), 6th Annual Report. 2008. Kisumu, Kenya; 2008.
18. INDEPTH. Population, Health and Survival at INDEPTH Sites. Ottawa Canada: International Development Research Centre; 2002.
19. Kenya Ministry of Public Health and Sanitation - Division of Leprosy Tuberculosis and Lung Disease. Annual Report. Nairobi; 2007.
20. van’t Hoog AH, Onyango J, Agaya J, Akeche G, Odero G, Lodenyo W, et al. Evaluation of TB and HIV services prior to introducing TB-HIV activities in two rural districts in western Kenya. Int J Tuberc Lung Dis. 2008; 12(3 Suppl 1): 32-8.
21. U.S. Department of State. The Power of Partnerships: The President’s Emergency Plan for AIDS Relief. 2008 Annual Report to Congress submitted by the Office of the U.S. Global AIDS Coordinator; January 2008.
22. Amolloh M, Medley A, Owuor P, Audi B, Sewe M, Muttai H, et al. Factors Associated with Early Uptake of HIV Care and Treatment Services after Testing HIV-Positive during Home-Based Testing and Counseling in Rural Western Kenya. 18th Conference on Retroviruses and Opportunistic Infections; 2011 February 27th – March 2nd 2011; Boston, MA, USA; 2011.
23. Lighter J, Rigaud M. Diagnosing childhood tuberculosis: traditional and innovative modalities. Curr Probl Pediatr Adolesc Health Care. 2009; 39(3): 61-88.
24. Measure DHS. Methodology of DHS Mortality Rates Estimation
DHS Statistics Live.
25. Dickman PW, Sloggett A, Hills M, Hakulinen T. Regression models for relative survival. Stat Med. 2004; 23(1): 51-64.
26. Dickman PW. Estimating and modelling relative survival using SAS. 2004 [cited 2011 10 April 2011]; Available from: http://www.pauldickman.com/rsmodel/
27. van’t Hoog AH, Marston BJ, Ayisi JG, Agaya JA, Muhenje O, Odeny LO, et al. Risk Factors for Inadequate TB case finding in Kenya: a Comparison of Prevalent and Self-reported TB Patients. in preparation. 2011.
28. Au-Yeung C, Kanters S, Ding E, Glaziou P, Anema A, Cooper CL, et al. Tuberculosis mortality in HIV-infected individuals: a cross-national systematic assessment. Clinical Epidemiology. 2011; 3: 21-9. 29. Borgdorff MW, Veen J, Kalisvaart
NA, Nagelkerke N. Mortality among tuberculosis patients in The Netherlands in the period 1993-1995. Eur Respir J. 1998; 11(4): 816-20.
30. Harries AD, Zachariah R, Corbett EL, Lawn SD, Santos-Filho ET, Chimzizi R, et al. The HIV-associated tuberculosis epidemic--when will we act? Lancet. 2010; 375(9729): 1906-19.
31. Franke MF, Robins JM, Mugabo J, Kaigamba F, Cain LE, Fleming JG, et al. Effectiveness of early antiretroviral therapy initiation to improve survival among HIV-infected adults with tuberculosis: a retrospective cohort study. PLoS Medicine. 2011; 8(5): e1001029. Epub 2011 May 3.
32. Abdool Karim SS, Naidoo K, Grobler A, Padayatchi N, Baxter C, Gray A, et al. Timing of initiation of antiretroviral drugs during tuberculosis therapy. N Engl J Med. 2010; 362(8): 697-706.
33. McIlleron H, Meintjes G, Burman WJ, Maartens G. Complications of antiretroviral therapy in patients with tuberculosis: drug interactions, toxicity, and immune reconstitution inflammatory syndrome. J Infect Dis. 2007; 196(Suppl 1): S63-75.
34. Sterling TR, Pham PA, Chaisson RE. HIV infection-related tuberculosis: clinical manifestations and treatment. Clin. 2010; 50(Suppl 3): S223-30.
4
35. Breen RA, Smith CJ, Cropley I, Johnson MA,Lipman MC. Does immune reconstitution syndrome promote active tuberculosis in patients receiving highly active antiretroviral therapy? Aids. 2005; 19(11): 1201-6.
36. Zachariah R, Harries AD, Manzi M, Gomani P, Teck R, Phillips M, et al. Acceptance of anti-retroviral therapy among patients infected with HIV and tuberculosis in rural Malawi is low and associated with cost of transport. PLoS One. 2006; 1: e121. 37. Zachariah R, Teck R, Ascurra O, Gomani
P, Manzi M, Humblet P, et al. Can we get more HIV-positive tuberculosis patients on antiretroviral treatment in a rural district of Malawi? Int J Tuberc Lung Dis. 2005; 9(3): 238-47.
38. Vree M, Huong NT, Duong BD, Sy DN, Van le N, Co NV, et al. Mortality and failure among tuberculosis patients who did not complete treatment in Vietnam: a cohort study. BMC Public Health. 2007; 7: 134. 39. Hargreaves NJ, Kadzakumanja O, Phiri
S, Nyangulu DS, Salaniponi FM, Harries AD, et al. What causes smear-negative pulmonary tuberculosis in Malawi, an area of high HIV seroprevalence? Int J Tuberc Lung Dis. 2001; 5(2): 113-22. 40. Cain KP, Anekthananon T, Burapat C,
Akksilp S, Mankhatitham W, Srinak C, et al. Causes of death in HIV-infected persons who have tuberculosis, Thailand. Emerg Infect Dis. 2009; 15(2): 258-64.
41. Nadol P, Stinson KW, Coggin W, Naicker M, Wells CD, Miller B, et al. Electronic tuberculosis surveillance systems: a tool for managing today’s TB programs. Int J Tuberc Lung Dis. 2008; 12(3 Suppl 1): 8-16. 42. Washington State Department of Health.
Guidelines for Using Confidence Intervals for Public Health Assessment. 2002.
