Tropical Medicine and International Health
doi:10.1111/tmi.12246
volume 19 no 3 pp 313–320 march 2014
Tuberculosis management time: an alternative parameter for measuring the tuberculosis infectious pool Solomon A. Yimer1,2,3,4, Carol Holm-Hansen1, Dag G. Storla5 and Gunnar A. Bjune2 1 2 3 4 5
Division of Infectious Disease Control, Norwegian Institute of Public Health, Oslo, Norway Institute of Health and Society, University of Oslo, Oslo, Norway Amhara Regional State Health Bureau, Bahir Dar, Ethiopia Department of Microbiology, Oslo University Hospital, Oslo, Norway Lovisenberg Diakonale Hospital, Oslo, Norway
Abstract
objective To demonstrate the application of TB management time as an alternative parameter to estimate the size of the tuberculosis infectious pool in West Gojjam Zone of Amhara Region, Ethiopia. methods In this study, we used the TB management time, i.e. the number of days from start of cough until start of treatment, to determine the infectious period. Patients with sputum smearpositive and smear-negative pulmonary TB, retreatment and an estimated number of undetected cases were included. The infectious pool was then estimated as the annual number of infectious person days in a defined population. results The TB management time of presently undiagnosed TB cases and sputum smear-positive patients contributed significantly to the infectious pool with 151 840 and 128 750 infectious person days per year, respectively. The total infectious pool including sputum smear-negative TB cases and retreatment patients in the study area was estimated at 325 410 person days or 15 447 person days per 100 000 population during the study year. conclusion Recording TB management time may be used to estimate the infectious pool of TB and to monitor programme performance in the community. keywords Tuberculosis, TB management time, diagnostic delay, treatment delay, case detection rate, Ethiopia
Introduction Tuberculosis (TB) remains a major public health challenge worldwide. Over 8.8 million new cases and 1.4 million deaths from TB occur annually (WHO 2011). The importance of TB control is addressed in the Millennium Development Goal (MDG) Six (UNDP 2012). The Stop TB Strategy is being implemented to achieve TB-related MDG targets and control the transmission of Mycobacterium tuberculosis (Mtb) in the community (WHO 2006). Measuring the size of the infectious pool is necessary to understand and monitor the trends of TB control programme performance. For more than two decades, the WHO-recommended case detection rate (CDR) was used as an indicator of the size of the infectious pool. CDR is defined as the notification rate of new smear-positive TB cases divided by the estimated incidence. WHO officially stopped reporting CDR estimates for smear-positive TB due to methodological problems (WHO 2011). Currently,
© 2014 John Wiley & Sons Ltd
WHO recommends using registered cases as a measure of TB burden (WHO 2011). WHO suggests that the best estimate for the burden of TB can be made by improving the quality of notification data and performing periodic prevalence surveys and data audits. However, even though this idea seems acceptable, it may not be practical given the poor capacity of most health facilities in highburden countries. Studies have shown that notification reports in many countries are not submitted in a timely manner and are often considered unreliable due to inaccuracy (Dye 2008; van der Wer et al. 2008; Glaziou et al. 2008). Lack of trained manpower is a serious challenge in many countries, and adequate training will be difficult to provide within the near future. The national CDR may not be relevant at the district level, and the psychological effect of not being able to meet the target is problematic. Conducting periodic prevalence surveys for most high TB burden countries is prohibitively expensive (Glaziou et al. 2008). 313
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
A model developed by Dye et al. (1998) provides a clear picture of TB transmission dynamics, illustrating how the different categories of patients with TB enter and exit the infectious pool. We previously modified this model and proposed a new simple parameter, i.e. using treatment delay as a key variable to measure the size of the TB infectious pool in a community (Storla et al. 2010). The proposed parameter is applied by defining the contribution of each category of patients with TB to the infectious pool. The contribution of each patient is estimated by measuring the length of delay from the start of cough until treatment is initiated (treatment delay). The parameter we proposed had potential limitations that needed further development (Storla et al. 2010). The size of the contributions from the different TB patient categories to the infectious pool had to be revised based on recent evidence, and the term ‘treatment delay’ did not accurately reflect the period of delay in question. To avoid confusion, it is important to find a term that can better describe the period from start of cough until the termination of infectiousness. The size of the infectious pool can be measured in terms of infectious person days. To address the limitations of our previous published work on this same topic, another study was conducted to investigate an improved application of the parameter. In the current study conducted in one area of Ethiopia, data from a cross-sectional study and unit TB registers were integrated, and the size of the infectious pool was estimated using TB management time given as infectious person days.
Methods Setting The study was conducted in West Gojjam Zone of Amhara Region, Ethiopia, with a population of 2 106 596 (CSA 2010). All TB diagnostic and treatment facilities in the study zone were included. Patients were enrolled in the study in a consecutive manner. Operational definitions of variables WHO definitions of different TB patient categories were used (WHO 2009). A new smear-positive or smear-negative TB case is ‘a patient who has never had treatment for TB or who has taken anti-TB drugs for less than 1 month’. Retreatment cases include treatment failure, relapse and treatment after default cases. A relapse case is a patient who reports back with smear-positive sputum after being declared cured of any form of TB following the completion of a full course of chemotherapy. 314
Treatment failure is a sputum smear-positive patient who remained smear-positive while on treatment or again became smear-positive 5 months or more after commencing treatment. A patient who was initially smear-negative before starting treatment and became smear-positive after the second month of treatment is also defined as treatment failure. A treatment after default case is a patient who has taken at least 4 weeks of treatment but has subsequently interrupted treatment for 2 months or more and returns to the health service with smear-positive sputum. TB management time is defined as the interval from the start of cough until treatment is initiated. The infectious pool model In this study, Dye’s dynamic model was modified to measure the size of the infectious pool of TB (Dye et al. 1998). As shown in Figure 1, the infectious pool comprises two groups of patients with TB: new TB cases (sputum smear-positive and sputum smear-negative) and retreatment cases. Each patient’s contribution to the infectious pool is equal to the number of days that the patient remains infectious, i.e. the period between the date stated by the patient as the onset of cough and the date the patient starts standard treatment. Our estimate of the infectious pool is thus based on a direct count of the days defining TB management time for each patient category multiplied by the number of patients treated in each category during 1 year, and by estimating the number of undiagnosed cases in the community. Extrapulmonary TB cases are considered not infectious in the model. Data collection To define the average TB management time for each patient category, we conducted a cross-sectional survey and reviewed unit TB registers in all public health TB diagnostic and treatment facilities in the study zone. The sample size for the cross-sectional survey was calculated using the formula required for estimating single proportions (Daniel 1987). Using a former study performed in Ethiopia that showed a 48% delay of more than 1 month (Yimer et al. 2005), a 95% CI and a margin error of 5%, the sample size was calculated to be 384. A pre-tested semi-structured questionnaire was used to collect data. The data included sociodemographic variables, type and duration of symptoms, date of first visit to medical facilities and date of first symptoms. A local calendar listing the main religious and national days was used to help patients remember the date of onset of symptoms. The data were collected by trained nurses and health officers, and data collection activity was regularly
© 2014 John Wiley & Sons Ltd
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
Non infected
Waned
Re-infected Reactivated
Infected
Detected and treated Self-cured
Active TB Smear-positive TB, smearnegative TB, retreatment cases
Figure 1 Model showing the dynamics of the TB infectious pool (modified from Dye et al. 1998).
Death
monitored by supervisors. Patient register cards, TB registration books and laboratory registries were crosschecked and validated to ensure the quality of data. Unit TB register data were reviewed to record the total number of smear-positive TB, smear-negative TB and retreatment cases seen at the study sites during the study period. TB diagnoses were made based on the treatment algorithm recommended by the National TB and Leprosy Control Program (MOH 2002). Ethics statement Ethical clearance was obtained from relevant authorities in Ethiopia and the Regional Committee for Medical Research Ethics in Eastern Norway (REK Øst). Informed consent was secured from all patients before the study was conducted. Data analysis Data were entered into statistical package for social sciences (SPSS) version 17 (SPSS Inc., Chicago, IL, USA). Before analysis was performed, data were cleaned for errors. Two individuals were involved in the data cleaning. One person read the data from the questionnaire, and the other looked at the computer screen to ensure
© 2014 John Wiley & Sons Ltd
that the data entered were identical to the information read from the questionnaire. Proportions, medians and interquartile range were computed after the data was cleaned. The median TB management time for smear-positive cases was calculated based on data from the cross-sectional study conducted in the study zone. Earlier studies indicated that an average of 2 weeks of treatment with standard combination of drugs resulted in rapid decline of infectiousness following which patients were considered noninfectious (Brooks et al. 1973; Rouillon et al. 1976). However, this assumption may not be correct. According to a recent study, 30% to 40% of smear-positive patients were still infectious after 2 weeks of standard therapy (Senkoro et al. 2010). Evidence indicates that an average time of 23–35 days is needed for sputum culture conversion after the commencement of standard treatment (Telzak et al. 1997; Dominguez-Castellano et al. 2003; Fitzwater et al. 2010; Parikh et al. 2012). Taking this into account, an average infectious period of 30 days was added to the observed median TB management time to estimate the total infectious period for smear-positive cases. Our estimate for the contribution of sputum smear-negative cases is based on the results of four studies. Results from the Netherlands, Canada and California showed that patients with smear-negative, 315
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
culture-positive TB were responsible for 13%, 17% and 22% of TB transmission relative to sputum smear-positive cases, respectively (Tostmann et al. 2008; Behr et al. 1999; Hernandez-Garduno et al. 2004). In a study from South Africa, more than 80% of TB/HIV cases with culture-confirmed TB were sputum smear-negative and were responsible for 25% of TB transmission (Lawn et al. 2009). These studies demonstrate that sputum smear-negative cases were responsible for transmission of TB ranging from 13% to 25%. In our estimate, an average value of 20% derived from the four studies was used. This value indicates that smear-negative TB cases were 20% as likely to transmit TB as smear-positive cases, and was applied to estimate the contribution of smear-negative TB to the infectious pool. TB management time and the total infectious period for the undiagnosed cases were calculated based on results from previous local studies in Ethiopia (Shargie et al. 2006; Zeleke & Fasil 2011) and a systematic review by Tiemersma et al. (2011). The systematic review indicated that undiagnosed patients with pulmonary TB (PTB) remain infectious for an average of 3 years (1095 days). It was estimated that 33% of patients with TB were undetected in Ethiopia (Shargie et al. 2006, Zeleke & Fasil 2012). Therefore, we used an average of 33% additional undiagnosed patients with a median TB management time of 365 days to calculate TB management time for the total number of estimated undiagnosed cases. The entire year was considered in the calculation as the infectious pool is calculated per year. The infectious period for retreatment cases (relapse, treatment failure and treatment after default) was estimated separately for each subcategory of patients. The treatment success rate of relapse cases when treated for category two was 76.4% in India (Mehra et al. 2008; Thomas et al. 2005) and 75% in Morocco (Ottmani et al. 2006), indicating a very good treatment outcome. It is conceivable that relapse cases report to health facilities quicker than other patients with TB, i.e. within 1 month of symptom onset, because they are familiar with TB symptoms and disease severity. Therefore, by adding a delay of 30 days in starting the retreatment regimen and an infectious period of 30 more days after retreatment start, a total infectious period of 60 days was estimated for relapse cases. The treatment success rate among category one failures when treated for category two was estimated at 75% (Harries et al. 2003) in Malawi. Studies from Nicaragua and Nepal indicated that the proportion of retreatment failure was 4.3% (Heldal et al. 2001) and 5% (Yoshiyama et al. 2010), respectively. These data indicate that in areas with low multidrug resistant TB (MDR-TB) and 316
successful TB control programme performance, the retreatment success rate is good. Patients with TB who remain smear-positive 5 months or more after commencing treatment are put on a Category 2 treatment regimen (WHO 2009). A majority of patients failed treatment at 5 months (Harries et al. 2003). Therefore, by adding an infectious period of 30 more days after retreatment start, the total infectious period for treatment failures was estimated at of 180 days. The infectious period for treatment after default cases was estimated by adding a median of 2 months treatment interruption period and a 1-month infectious period after retreatment start. The total infectious period for this group of patients was estimated at 90 days. However, in areas where efficient TB control programmes are established, the number of treatment after default cases will be very low and the contribution to the infectious pool from these patients will be negligible. Data on the total number of sputum smear-positive, sputum smear-negative and retreatment TB cases treated in the study year was obtained from the unit TB registers. The total infectious person days for each patient category were calculated by multiplying the total number of TB patients registered in each category during 1 year by the estimated average infectious period for each category of patients with TB. The infectious pool was measured in terms of person days/100 000 population per year. The total number of infectious person days and an estimate of the infectious pool were calculated by combining the total number of infectious person days contributed by each patient category in the study zone during 1 year. The infectious pool calculations were based on the following assumptions and equation. Assumption: let median infectious period and total number (seen during 1 year) for smear-positives, smearnegatives, relapse cases, treatment failures, and undiagnosed cases be A1 and N1, A2 and N2, A3 and N3, A4 and N4, A5 and N5, A6 and N6 and A7 with N7, respectively. Hence, the estimated total infectious pool can be calculated using the following equation: Total infectious pool ¼ A1 Niþ...þ A7 N7þ ¼
7 X
A i Ni
i¼1
Results The median TB management time for new smear-positive cases was 73 days (interquartile range, 45–150 days). By adding an infectious period of 30 days after the start of standard treatment, the contribution of one smear-positive case to the infectious pool was estimated at 103 infectious person days. A total of 1250 new
© 2014 John Wiley & Sons Ltd
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
smear-positive patients who were treated during the study year yielded 128 750 infectious person days (Table 1). There were 1998 new sputum smear-negative cases treated during the study year. It was estimated that smear-negative cases constituted approximately 20% of the transmission rate relative to smear-positive cases or 20 infectious days per case. Thus, the smear-negative cases contributed a total of 39 960 infectious person days. It was estimated that approximately 1/3 of all new sputum smear-positive cases was undiagnosed or not notified in the study zone. The estimated median infectious period of each untreated case was 365 days yielding a total of 151 840 infectious person days for the 416 cases. Forty-five relapse cases registered during the year with an infectious period estimated at 60 days contributed 2700 infectious person days. The estimated median infectious period for each treatment failure case was 180 days. The nine treatment failure cases contributed a total of 1620 infectious person days. The median infectious period for treatment after default was 90 days. The six treatment after default cases contributed 540 infectious person days. Based on the above calculations, the total infectious pool for West Gojjam Zone of Amhara Region (population 2 106 596) was estimated at 325 410 person days or 15 447 person days/100 000 population. Discussion Measuring the burden of TB is essential to understand transmission dynamics in the community. In this study, we applied a simple alternative tool to estimate the size of the infectious pool of TB. The tool uses the infectious periods of pulmonary TB (PTB) cases as an important
parameter to measure the size of the infectious pool. Defining the average duration of infectiousness for each patient category is crucial for practical application of the tool. New smear-positive and smear-negative PTB cases comprise the major portion of the infectious pool. To measure the infectious period, we collected data from a cross-sectional survey and unit TB register books, and applied previous study findings. No study to date has precisely determined the exact time at which a patient with active PTB becomes non-infectious after the initiation anti-TB treatment. The combination of various methods applied to estimate the infectious period in this study may provide a reasonable estimate of the relative contribution of new PTB cases to the total infectious pool of TB in the study zone. Estimating the number of undiagnosed cases in the community is a challenge. It is necessary to determine whether these cases are patients who are ‘not yet detected’ or ‘never detected’. A randomised intervention control trial addressing active case finding was conducted in Ethiopia during 2006 in an area where the directly observed treatment short course (DOTS) coverage was 75% (Shargie et al. 2006). The ratio of sputum smear-positive cases already on treatment to newly detected sputum smear-positive cases was 2:1, indicating that for every two smear-positive TB cases identified at a health facility there was one undetected case in the community. The proportion of smear-positive cases detected in the community was 1/3 of the cases detected at DOTS facilities. Thus, the majority of cases were already diagnosed at DOTS centres before active case finding was conducted. Repeated active case finding was conducted in the same population to identify additional cases. While there was no increase in CDR, diagnostic delay was reduced. This suggests that the undiagnosed
Table 1 Estimated infectious pool of TB in West Gojjam Zone of Amhara Region, Ethiopia, in 2009
TB cases
Category
Registered
Smear-positive Smear-negatives Retreatment
Not Registered
Sub category
Estimated median infectious period in days
Total number patients in a year
Relapse Treatment failure Treatment after default Undetected cases
103 20 60 180 90 365
1250 1998 45 9 6 416*
Total
Total estimated infectious person days in a year 128 750 39 960 2700 1620 540 151 840 325 410 person days or 15 447 person days/100 000 population
*Undiagnosed (undetected cases) (2:1).
© 2014 John Wiley & Sons Ltd
317
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
TB cases were ‘not yet detected’ rather than ‘never detected’. A recent national prevalence survey in Ethiopia reported a prevalence rate of 105/100 000 population for smearpositive TB with a corresponding CDR of 71% (Zeleke & Fasil 2011). The current WHO CDR estimate for all forms of TB in Ethiopia is 72% (CI 66–78) (WHO 2011). DOTS population coverage in the country and in the study region is now over 95%, indicating that the majority of active TB cases have access to TB diagnosis and treatment. It is thus likely that the undetected TB cases will be identified sooner or later, i.e. the undiagnosed cases are primarily ‘not yet detected’ and in the process of reaching a diagnosis. Based on the results of the 2006 intervention control trial data and the 2011 national prevalence survey, an average of 33% additional undiagnosed patients was used as an estimate of the proportion of the undiagnosed group in our study. This proportion is a reasonable estimate for the undiagnosed TB cases. In our previous work, treatment delay was a key variable to define the period from the start of symptoms until treatment was initiated. We have replaced this variable with TB management time in this study because the term treatment delay may be misunderstood. Treatment delay is often perceived as the period from the time of diagnosis until the start of treatment. A more appropriate term is needed to describe the infectious period that is the basis for the size of the pool. The proposed new parameter TB management time better defines the infectious period from the start of cough until treatment is initiated. This term refers to the time required for the patient to manage the symptoms, i.e. the patient’s actions in relation to health-seeking activities (patient delay). It also includes the time used by the health facility to manage the patient’s symptoms, i.e. providing timely diagnosis and treatment (diagnostic and treatment delay). The use of TB management time reduces possible confusion by including all delay periods, patient delay, diagnostic delay and treatment delay, in addition to the infectious period after the initiation of treatment (Figure 2).
Defining the onset of TB is crucial when estimating the infectious period. In this study, the date of cough onset is taken as the point of entry into the infectious pool. Cough is the first symptom among approximately 95% of patients with active TB and is the most important variable in TB transmission (Yimer et al. 2005; Lawn et al. 1998; Demissie et al. 2002). We propose that the infectious period is the time during which the patient coughs, and that TB management time defines each patient’s contribution to the infectious pool. Cough is a noticeable symptom, and the time of onset may be recalled without difficulty by most patients. The dates of cough onset and start of treatment are important parameters that define TB management time and can be obtained by simply including an additional question on the unit TB registration forms. The application of the tool can be monitored at the implementing local health facility level by examining the total infectious pool over a specified period of time. A decrease or increase in the infectious pool may indicate changes in the performance of the TB control programme at that level. At present, there is no adequate system for monitoring the infectious pool and TB control programme performance. An alternative method that is simple, inexpensive and can be implemented at the local level is urgently needed to monitor the infectious pool and trends in TB control programme performance. TB management time is a new concept for monitoring the infectious pool of TB in a population. Limitations of the study There are a number of methodological concerns that should be considered. Cough is used as marker of infectiousness in this study. It is expected that some patients may have difficulty in remembering the exact date of cough onset due to longer durations and may therefore be subject to a recall bias. The relative number of undiagnosed cases in different settings may differ. The proportion of the undiagnosed
Patient health seeking and diagnosis at GHF/PHF or others
Cough (TB symptoms)
Patient delay
Diagnosis
Diagnostic delay
Treatment start
Treatment delay
TB management time Figure 2 Relation between different delay periods among patients with TB. GHF: Government health facility; PHF: Private health facility.
318
© 2014 John Wiley & Sons Ltd
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
group used in this article may not be the true figure for our study population for two reasons. Firstly, the local prevalence study that we used to derive our estimate was conducted in 2006 and may not accurately indicate the situation of undiagnosed TB in the study area. Secondly, the size of undetected smear-negative TB has not been estimated; we may thus have underestimated the total size of undiagnosed PTB cases in the study area. This parameter should ideally be measured in a survey of the same population before calculating the infectious pool and full-scale implementation of the tool. The average sputum and culture conversion time used to estimate the infectious period after start of TB treatment was based on previous studies from other countries. This may have resulted in under- or overestimation of the infectious period in our study. More studies addressing the time and factors of smear and culture conversions are needed. The fact that culture was not performed may have resulted in over diagnoses specifically among sputum smear-negative patients, and this may have resulted in overestimation of the size of infectious pool. Conclusions We used a systematic registration of TB management time in unit TB registry books to estimate the infectious pool of TB and determine trends of programme performance at the local level. Additional prospective local studies need to be conducted, and TB management time should be validated as a surrogate marker for programme performance before full-scale implementation of the tool. Potential limitations of this tool must be addressed to secure optimal application. The patient’s ability to recall the duration of symptoms, the size of the undiagnosed group, the contribution of smear-negative cases particularly in high TB burden countries and the contribution of retreatment cases including MDR-TB must be determined to accurately estimate the size of the infectious pool at local level. More research is warranted to develop an acceptable, simple and inexpensive tool to monitor the infectious pool of TB in resource poor settings. Acknowledgements We are grateful to the Research Council of Norway, GLOBVAC Programme (grant 192468/S50), the University of Oslo, the Amhara Regional State Health Bureau and West Gojjam Zone Health Department for funding and facilitating this study. The authors thank Bjørn Haneberg for constructive discussions.
© 2014 John Wiley & Sons Ltd
References Behr MA, Warren SA, Salamon H, et al. (1999) Transmission of Mycobacterium tuberculosis from patients smear-negative for acid-fast bacilli. Lancet 353, 444–449. Brooks SM, Lassiter NL & Young EC (1973) A pilot study concerning the infection risk of sputum positive tuberculosis patients on chemotherapy. The American review of respiratory disease 108, 799–804. CSA (2010) Statistical Tables for the 2007 Population and Housing Census of Ethiopia. CSA, Addis Ababa. Daniel W (1987) Biostatistics: A foundation for analysis in the health sciences, 4th edn. John Wiley & Sons Inc, New York, pp. 155–156. Demissie M, Lindtjørn B & Berhane Y (2002) Patient and health service delay in the diagnosis of pulmonary tuberculosis in Ethiopia. BMC Public Health 2, 23. Dominguez-Castellano A, Muniain MA, Rodriguez-Bano J, et al. (2003) Factors associated with time to sputum smear conversion in active pulmonary tuberculosis. International Journal of Tuberculosis and Lung Disease 7, 432–438. Dye C (2008) Breaking a law: Tuberculosis disobeys Styblo’s rule. Bulletin of the World Health Organization 86, 1. Dye C, Garnett GP, Sleeman K & Williams BG (1998) Prospects for worldwide tuberculosis control under the WHO DOTS strategy. Directly observed short-course therapy. The Lancet 352, 1886–1891. Fitzwater SP, Caviedes L, Gilman RH, et al. (2010) Prolonged infectiousness of tuberculosis patients in a directly observed therapy short-course program with standardized therapy. Clinical Infectious Diseases 51, 371–378. Glaziou P, van der Werf OI & Dye C (2008) Tuberculosis prevalence surveys: rationale and cost. International Journal of Tuberculosis and Lung Disease 12, 1003–1008. Harries AD, Nyirenda TE, Kemp JR, Squire BS, Godfrey-Faussett P & Salaniponi FM, (2003) Management and outcome of tuberculosis patients who fail treatment under routine programme conditions in Malawi. International Journal of Tuberculosis and Lung Disease 11, 1040–1044. Heldal E, Arnadottir T, Cruz JR, Tardencilla A & Chacon L (2001) Low failure rate in standardised retreatment of tuberculosis in Nicaragua: patient category, drug resistance and survival of ‘chronic’ patients. International Journal of Tuberculosis and Lung Disease 5, 129–136. Hernandez-Garduno CV, Kunimoto D, Elwood RK, Black WA & FitzGerald JM (2004) Transmission of tuberculosis from smear negative patients: a molecular epidemiology study. Thorax 59, 286–290. Lawn SD, Afful B & Acheampong JW (1998) Pulmonary tuberculosis: diagnostic delay in Ghanaian adults. International Journal of Tuberculosis and Lung Disease 2, 635–640. Lawn SD, Edwards D & Wood R (2009) Tuberculosis transmission from patients with smear-negative pulmonary tuberculosis in sub-Saharan Africa. Clinical Infectious Diseases 48, 496–497. Mehra RK, Dhingra VK, Nish A & Vashist RP (2008) Study of relapse and failure cases of CAT I retreated with CAT II under
319
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
RNTCP–an eleven year follow up. The Indian Journal of Tuberculosis 55, 188–191. MOH (2002) National Tuberculosis and Leprosy Prevention and Control Program, Manual. MOH , Addis Ababa, Ethiopia. Ottmani SE, Zignol M, Bencheikh N, La^asri L, Chaouki N & Mahjour J (2006) Results of cohort analysis by category of tuberculosis retreatment cases in Morocco from 1996 to 2003. International Journal of Tuberculosis and Lung Disease 10, 1367–1372. Parikh R, Nataraj G, Kanade S, Khatri V & Mehta P (2012) Time to sputum conversion in smear positive pulmonary TB patients on category I DOTS and factors delaying it. Journal of the Association of Physicians of India 60, 22–26. Rouillon A, Perdrizet S & Parrot R (1976) Transmission of tubercle bacilli: the effects of chemotherapy. Tubercle 57, 275–299. Senkoro M, Mfinanga SG & Mørkve O (2010) Smear microscopy and culture conversion rates among smear positive pulmonary tuberculosis patients by HIV status in Dar es Salaam, Tanzania. BMC infectious diseases 16(10), 210. Shargie EB, Mørkve O & Lindtjørn B (2006) Tuberculosis casefinding through a village outreach programme in a rural setting in southern Ethiopia: community randomized trial. Bulletin of the World Health Organization 84, 112–119. Storla DG, Yimer S & Bjune GA (2010) Can treatment delay be utilized as a key variable for monitoring the pool of infectious tuberculosis in a population? The Journal of Infection in Developing Countries 4, 83–90. Telzak EE, Fazal BA, Pollard CL, Turett GS, Justman JE & Blum S (1997) Factors influencing time to sputum conversion among patients with smear-positive pulmonary tuberculosis. Clinical Infectious Diseases 25, 666–670. Thomas A, Gopi PG, Santha T, et al. (2005) Predictors of relapse among pulmonary tuberculosis patients treated in a
DOTS programme in South India. International Journal of Tuberculosis and Lung Disease 9, 556–561. Tiemersma EW, van der Werf MJ, Borgdorff MW, Williams BG & Nagelkerke NJ (2011) Natural history of tuberculosis: duration and fatality of untreated pulmonary tuberculosis in HIV negative patients: a systematic review. PLoS ONE 6, e17601. Tostmann A, Kik SV, Kalisvaart NA, et al. (2008) Tuberculosis transmission by patients with smear-negative pulmonary tuberculosis in a large cohort in the Netherlands. Clinical Infectious Diseases 47, 1135–1142. UNDP (2012) United Nations, Millennium Development Goals Report. UNDP, New York, pp 44–45. van der Wer MJ, Enarson DA & Borgdorff MW (2008) How to identify tuberculosis cases in a prevalence survey. International Journal of Tuberculosis and Lung Disease 12, 1255–1260. WHO (2006) The Stop TB strategy. WHO/HTM/TB/2006.368, WHO, Geneva. WHO (2009) Management of tuberculosis training for health facility staff, 2nd edn. WHO/HTM/TB/2009.423c, WHO, Geneva. WHO (2011) Global tuberculosis control: WHO Annual Report (WHO/HTM/TB/2012.6), WHO, Geneva. Yimer S, Bjune G & Alene G (2005) Diagnostic and treatment delay among pulmonary tuberculosis patients in Ethiopia A cross-sectional study. BMC Infectious Diseases, 5, 112. Yoshiyama T, Shrestha B & Maharjan B (2010) Risk of relapse and failure after retreatment with the Category II regimen in Nepal. International Journal of Tuberculosis and Lung Disease 14, 1418–1423. Zeleke A & Fasil DAK (2011) Ethiopian national tuberculosis prevalence survey 2010–2011: Preliminary Result: Paper presented at the 42nd Union World Conference on Lung Health; 2011 Oct 26–30; Lille, France.
Corresponding Author Solomon A Yimer, Division of Infectious Disease Control, Department of Bacteriology and Immunology, Norwegian Institute of Public Health, PO Box 4404 Nydalen, 0403 Oslo, Norway. Tel.: +47 47687670; +47 21077000; Fax: +47 22353605; E-mail: [email protected]
320
© 2014 John Wiley & Sons Ltd
doi:10.1111/tmi.12246
volume 19 no 3 pp 313–320 march 2014
Tuberculosis management time: an alternative parameter for measuring the tuberculosis infectious pool Solomon A. Yimer1,2,3,4, Carol Holm-Hansen1, Dag G. Storla5 and Gunnar A. Bjune2 1 2 3 4 5
Division of Infectious Disease Control, Norwegian Institute of Public Health, Oslo, Norway Institute of Health and Society, University of Oslo, Oslo, Norway Amhara Regional State Health Bureau, Bahir Dar, Ethiopia Department of Microbiology, Oslo University Hospital, Oslo, Norway Lovisenberg Diakonale Hospital, Oslo, Norway
Abstract
objective To demonstrate the application of TB management time as an alternative parameter to estimate the size of the tuberculosis infectious pool in West Gojjam Zone of Amhara Region, Ethiopia. methods In this study, we used the TB management time, i.e. the number of days from start of cough until start of treatment, to determine the infectious period. Patients with sputum smearpositive and smear-negative pulmonary TB, retreatment and an estimated number of undetected cases were included. The infectious pool was then estimated as the annual number of infectious person days in a defined population. results The TB management time of presently undiagnosed TB cases and sputum smear-positive patients contributed significantly to the infectious pool with 151 840 and 128 750 infectious person days per year, respectively. The total infectious pool including sputum smear-negative TB cases and retreatment patients in the study area was estimated at 325 410 person days or 15 447 person days per 100 000 population during the study year. conclusion Recording TB management time may be used to estimate the infectious pool of TB and to monitor programme performance in the community. keywords Tuberculosis, TB management time, diagnostic delay, treatment delay, case detection rate, Ethiopia
Introduction Tuberculosis (TB) remains a major public health challenge worldwide. Over 8.8 million new cases and 1.4 million deaths from TB occur annually (WHO 2011). The importance of TB control is addressed in the Millennium Development Goal (MDG) Six (UNDP 2012). The Stop TB Strategy is being implemented to achieve TB-related MDG targets and control the transmission of Mycobacterium tuberculosis (Mtb) in the community (WHO 2006). Measuring the size of the infectious pool is necessary to understand and monitor the trends of TB control programme performance. For more than two decades, the WHO-recommended case detection rate (CDR) was used as an indicator of the size of the infectious pool. CDR is defined as the notification rate of new smear-positive TB cases divided by the estimated incidence. WHO officially stopped reporting CDR estimates for smear-positive TB due to methodological problems (WHO 2011). Currently,
© 2014 John Wiley & Sons Ltd
WHO recommends using registered cases as a measure of TB burden (WHO 2011). WHO suggests that the best estimate for the burden of TB can be made by improving the quality of notification data and performing periodic prevalence surveys and data audits. However, even though this idea seems acceptable, it may not be practical given the poor capacity of most health facilities in highburden countries. Studies have shown that notification reports in many countries are not submitted in a timely manner and are often considered unreliable due to inaccuracy (Dye 2008; van der Wer et al. 2008; Glaziou et al. 2008). Lack of trained manpower is a serious challenge in many countries, and adequate training will be difficult to provide within the near future. The national CDR may not be relevant at the district level, and the psychological effect of not being able to meet the target is problematic. Conducting periodic prevalence surveys for most high TB burden countries is prohibitively expensive (Glaziou et al. 2008). 313
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
A model developed by Dye et al. (1998) provides a clear picture of TB transmission dynamics, illustrating how the different categories of patients with TB enter and exit the infectious pool. We previously modified this model and proposed a new simple parameter, i.e. using treatment delay as a key variable to measure the size of the TB infectious pool in a community (Storla et al. 2010). The proposed parameter is applied by defining the contribution of each category of patients with TB to the infectious pool. The contribution of each patient is estimated by measuring the length of delay from the start of cough until treatment is initiated (treatment delay). The parameter we proposed had potential limitations that needed further development (Storla et al. 2010). The size of the contributions from the different TB patient categories to the infectious pool had to be revised based on recent evidence, and the term ‘treatment delay’ did not accurately reflect the period of delay in question. To avoid confusion, it is important to find a term that can better describe the period from start of cough until the termination of infectiousness. The size of the infectious pool can be measured in terms of infectious person days. To address the limitations of our previous published work on this same topic, another study was conducted to investigate an improved application of the parameter. In the current study conducted in one area of Ethiopia, data from a cross-sectional study and unit TB registers were integrated, and the size of the infectious pool was estimated using TB management time given as infectious person days.
Methods Setting The study was conducted in West Gojjam Zone of Amhara Region, Ethiopia, with a population of 2 106 596 (CSA 2010). All TB diagnostic and treatment facilities in the study zone were included. Patients were enrolled in the study in a consecutive manner. Operational definitions of variables WHO definitions of different TB patient categories were used (WHO 2009). A new smear-positive or smear-negative TB case is ‘a patient who has never had treatment for TB or who has taken anti-TB drugs for less than 1 month’. Retreatment cases include treatment failure, relapse and treatment after default cases. A relapse case is a patient who reports back with smear-positive sputum after being declared cured of any form of TB following the completion of a full course of chemotherapy. 314
Treatment failure is a sputum smear-positive patient who remained smear-positive while on treatment or again became smear-positive 5 months or more after commencing treatment. A patient who was initially smear-negative before starting treatment and became smear-positive after the second month of treatment is also defined as treatment failure. A treatment after default case is a patient who has taken at least 4 weeks of treatment but has subsequently interrupted treatment for 2 months or more and returns to the health service with smear-positive sputum. TB management time is defined as the interval from the start of cough until treatment is initiated. The infectious pool model In this study, Dye’s dynamic model was modified to measure the size of the infectious pool of TB (Dye et al. 1998). As shown in Figure 1, the infectious pool comprises two groups of patients with TB: new TB cases (sputum smear-positive and sputum smear-negative) and retreatment cases. Each patient’s contribution to the infectious pool is equal to the number of days that the patient remains infectious, i.e. the period between the date stated by the patient as the onset of cough and the date the patient starts standard treatment. Our estimate of the infectious pool is thus based on a direct count of the days defining TB management time for each patient category multiplied by the number of patients treated in each category during 1 year, and by estimating the number of undiagnosed cases in the community. Extrapulmonary TB cases are considered not infectious in the model. Data collection To define the average TB management time for each patient category, we conducted a cross-sectional survey and reviewed unit TB registers in all public health TB diagnostic and treatment facilities in the study zone. The sample size for the cross-sectional survey was calculated using the formula required for estimating single proportions (Daniel 1987). Using a former study performed in Ethiopia that showed a 48% delay of more than 1 month (Yimer et al. 2005), a 95% CI and a margin error of 5%, the sample size was calculated to be 384. A pre-tested semi-structured questionnaire was used to collect data. The data included sociodemographic variables, type and duration of symptoms, date of first visit to medical facilities and date of first symptoms. A local calendar listing the main religious and national days was used to help patients remember the date of onset of symptoms. The data were collected by trained nurses and health officers, and data collection activity was regularly
© 2014 John Wiley & Sons Ltd
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
Non infected
Waned
Re-infected Reactivated
Infected
Detected and treated Self-cured
Active TB Smear-positive TB, smearnegative TB, retreatment cases
Figure 1 Model showing the dynamics of the TB infectious pool (modified from Dye et al. 1998).
Death
monitored by supervisors. Patient register cards, TB registration books and laboratory registries were crosschecked and validated to ensure the quality of data. Unit TB register data were reviewed to record the total number of smear-positive TB, smear-negative TB and retreatment cases seen at the study sites during the study period. TB diagnoses were made based on the treatment algorithm recommended by the National TB and Leprosy Control Program (MOH 2002). Ethics statement Ethical clearance was obtained from relevant authorities in Ethiopia and the Regional Committee for Medical Research Ethics in Eastern Norway (REK Øst). Informed consent was secured from all patients before the study was conducted. Data analysis Data were entered into statistical package for social sciences (SPSS) version 17 (SPSS Inc., Chicago, IL, USA). Before analysis was performed, data were cleaned for errors. Two individuals were involved in the data cleaning. One person read the data from the questionnaire, and the other looked at the computer screen to ensure
© 2014 John Wiley & Sons Ltd
that the data entered were identical to the information read from the questionnaire. Proportions, medians and interquartile range were computed after the data was cleaned. The median TB management time for smear-positive cases was calculated based on data from the cross-sectional study conducted in the study zone. Earlier studies indicated that an average of 2 weeks of treatment with standard combination of drugs resulted in rapid decline of infectiousness following which patients were considered noninfectious (Brooks et al. 1973; Rouillon et al. 1976). However, this assumption may not be correct. According to a recent study, 30% to 40% of smear-positive patients were still infectious after 2 weeks of standard therapy (Senkoro et al. 2010). Evidence indicates that an average time of 23–35 days is needed for sputum culture conversion after the commencement of standard treatment (Telzak et al. 1997; Dominguez-Castellano et al. 2003; Fitzwater et al. 2010; Parikh et al. 2012). Taking this into account, an average infectious period of 30 days was added to the observed median TB management time to estimate the total infectious period for smear-positive cases. Our estimate for the contribution of sputum smear-negative cases is based on the results of four studies. Results from the Netherlands, Canada and California showed that patients with smear-negative, 315
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
culture-positive TB were responsible for 13%, 17% and 22% of TB transmission relative to sputum smear-positive cases, respectively (Tostmann et al. 2008; Behr et al. 1999; Hernandez-Garduno et al. 2004). In a study from South Africa, more than 80% of TB/HIV cases with culture-confirmed TB were sputum smear-negative and were responsible for 25% of TB transmission (Lawn et al. 2009). These studies demonstrate that sputum smear-negative cases were responsible for transmission of TB ranging from 13% to 25%. In our estimate, an average value of 20% derived from the four studies was used. This value indicates that smear-negative TB cases were 20% as likely to transmit TB as smear-positive cases, and was applied to estimate the contribution of smear-negative TB to the infectious pool. TB management time and the total infectious period for the undiagnosed cases were calculated based on results from previous local studies in Ethiopia (Shargie et al. 2006; Zeleke & Fasil 2011) and a systematic review by Tiemersma et al. (2011). The systematic review indicated that undiagnosed patients with pulmonary TB (PTB) remain infectious for an average of 3 years (1095 days). It was estimated that 33% of patients with TB were undetected in Ethiopia (Shargie et al. 2006, Zeleke & Fasil 2012). Therefore, we used an average of 33% additional undiagnosed patients with a median TB management time of 365 days to calculate TB management time for the total number of estimated undiagnosed cases. The entire year was considered in the calculation as the infectious pool is calculated per year. The infectious period for retreatment cases (relapse, treatment failure and treatment after default) was estimated separately for each subcategory of patients. The treatment success rate of relapse cases when treated for category two was 76.4% in India (Mehra et al. 2008; Thomas et al. 2005) and 75% in Morocco (Ottmani et al. 2006), indicating a very good treatment outcome. It is conceivable that relapse cases report to health facilities quicker than other patients with TB, i.e. within 1 month of symptom onset, because they are familiar with TB symptoms and disease severity. Therefore, by adding a delay of 30 days in starting the retreatment regimen and an infectious period of 30 more days after retreatment start, a total infectious period of 60 days was estimated for relapse cases. The treatment success rate among category one failures when treated for category two was estimated at 75% (Harries et al. 2003) in Malawi. Studies from Nicaragua and Nepal indicated that the proportion of retreatment failure was 4.3% (Heldal et al. 2001) and 5% (Yoshiyama et al. 2010), respectively. These data indicate that in areas with low multidrug resistant TB (MDR-TB) and 316
successful TB control programme performance, the retreatment success rate is good. Patients with TB who remain smear-positive 5 months or more after commencing treatment are put on a Category 2 treatment regimen (WHO 2009). A majority of patients failed treatment at 5 months (Harries et al. 2003). Therefore, by adding an infectious period of 30 more days after retreatment start, the total infectious period for treatment failures was estimated at of 180 days. The infectious period for treatment after default cases was estimated by adding a median of 2 months treatment interruption period and a 1-month infectious period after retreatment start. The total infectious period for this group of patients was estimated at 90 days. However, in areas where efficient TB control programmes are established, the number of treatment after default cases will be very low and the contribution to the infectious pool from these patients will be negligible. Data on the total number of sputum smear-positive, sputum smear-negative and retreatment TB cases treated in the study year was obtained from the unit TB registers. The total infectious person days for each patient category were calculated by multiplying the total number of TB patients registered in each category during 1 year by the estimated average infectious period for each category of patients with TB. The infectious pool was measured in terms of person days/100 000 population per year. The total number of infectious person days and an estimate of the infectious pool were calculated by combining the total number of infectious person days contributed by each patient category in the study zone during 1 year. The infectious pool calculations were based on the following assumptions and equation. Assumption: let median infectious period and total number (seen during 1 year) for smear-positives, smearnegatives, relapse cases, treatment failures, and undiagnosed cases be A1 and N1, A2 and N2, A3 and N3, A4 and N4, A5 and N5, A6 and N6 and A7 with N7, respectively. Hence, the estimated total infectious pool can be calculated using the following equation: Total infectious pool ¼ A1 Niþ...þ A7 N7þ ¼
7 X
A i Ni
i¼1
Results The median TB management time for new smear-positive cases was 73 days (interquartile range, 45–150 days). By adding an infectious period of 30 days after the start of standard treatment, the contribution of one smear-positive case to the infectious pool was estimated at 103 infectious person days. A total of 1250 new
© 2014 John Wiley & Sons Ltd
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
smear-positive patients who were treated during the study year yielded 128 750 infectious person days (Table 1). There were 1998 new sputum smear-negative cases treated during the study year. It was estimated that smear-negative cases constituted approximately 20% of the transmission rate relative to smear-positive cases or 20 infectious days per case. Thus, the smear-negative cases contributed a total of 39 960 infectious person days. It was estimated that approximately 1/3 of all new sputum smear-positive cases was undiagnosed or not notified in the study zone. The estimated median infectious period of each untreated case was 365 days yielding a total of 151 840 infectious person days for the 416 cases. Forty-five relapse cases registered during the year with an infectious period estimated at 60 days contributed 2700 infectious person days. The estimated median infectious period for each treatment failure case was 180 days. The nine treatment failure cases contributed a total of 1620 infectious person days. The median infectious period for treatment after default was 90 days. The six treatment after default cases contributed 540 infectious person days. Based on the above calculations, the total infectious pool for West Gojjam Zone of Amhara Region (population 2 106 596) was estimated at 325 410 person days or 15 447 person days/100 000 population. Discussion Measuring the burden of TB is essential to understand transmission dynamics in the community. In this study, we applied a simple alternative tool to estimate the size of the infectious pool of TB. The tool uses the infectious periods of pulmonary TB (PTB) cases as an important
parameter to measure the size of the infectious pool. Defining the average duration of infectiousness for each patient category is crucial for practical application of the tool. New smear-positive and smear-negative PTB cases comprise the major portion of the infectious pool. To measure the infectious period, we collected data from a cross-sectional survey and unit TB register books, and applied previous study findings. No study to date has precisely determined the exact time at which a patient with active PTB becomes non-infectious after the initiation anti-TB treatment. The combination of various methods applied to estimate the infectious period in this study may provide a reasonable estimate of the relative contribution of new PTB cases to the total infectious pool of TB in the study zone. Estimating the number of undiagnosed cases in the community is a challenge. It is necessary to determine whether these cases are patients who are ‘not yet detected’ or ‘never detected’. A randomised intervention control trial addressing active case finding was conducted in Ethiopia during 2006 in an area where the directly observed treatment short course (DOTS) coverage was 75% (Shargie et al. 2006). The ratio of sputum smear-positive cases already on treatment to newly detected sputum smear-positive cases was 2:1, indicating that for every two smear-positive TB cases identified at a health facility there was one undetected case in the community. The proportion of smear-positive cases detected in the community was 1/3 of the cases detected at DOTS facilities. Thus, the majority of cases were already diagnosed at DOTS centres before active case finding was conducted. Repeated active case finding was conducted in the same population to identify additional cases. While there was no increase in CDR, diagnostic delay was reduced. This suggests that the undiagnosed
Table 1 Estimated infectious pool of TB in West Gojjam Zone of Amhara Region, Ethiopia, in 2009
TB cases
Category
Registered
Smear-positive Smear-negatives Retreatment
Not Registered
Sub category
Estimated median infectious period in days
Total number patients in a year
Relapse Treatment failure Treatment after default Undetected cases
103 20 60 180 90 365
1250 1998 45 9 6 416*
Total
Total estimated infectious person days in a year 128 750 39 960 2700 1620 540 151 840 325 410 person days or 15 447 person days/100 000 population
*Undiagnosed (undetected cases) (2:1).
© 2014 John Wiley & Sons Ltd
317
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
TB cases were ‘not yet detected’ rather than ‘never detected’. A recent national prevalence survey in Ethiopia reported a prevalence rate of 105/100 000 population for smearpositive TB with a corresponding CDR of 71% (Zeleke & Fasil 2011). The current WHO CDR estimate for all forms of TB in Ethiopia is 72% (CI 66–78) (WHO 2011). DOTS population coverage in the country and in the study region is now over 95%, indicating that the majority of active TB cases have access to TB diagnosis and treatment. It is thus likely that the undetected TB cases will be identified sooner or later, i.e. the undiagnosed cases are primarily ‘not yet detected’ and in the process of reaching a diagnosis. Based on the results of the 2006 intervention control trial data and the 2011 national prevalence survey, an average of 33% additional undiagnosed patients was used as an estimate of the proportion of the undiagnosed group in our study. This proportion is a reasonable estimate for the undiagnosed TB cases. In our previous work, treatment delay was a key variable to define the period from the start of symptoms until treatment was initiated. We have replaced this variable with TB management time in this study because the term treatment delay may be misunderstood. Treatment delay is often perceived as the period from the time of diagnosis until the start of treatment. A more appropriate term is needed to describe the infectious period that is the basis for the size of the pool. The proposed new parameter TB management time better defines the infectious period from the start of cough until treatment is initiated. This term refers to the time required for the patient to manage the symptoms, i.e. the patient’s actions in relation to health-seeking activities (patient delay). It also includes the time used by the health facility to manage the patient’s symptoms, i.e. providing timely diagnosis and treatment (diagnostic and treatment delay). The use of TB management time reduces possible confusion by including all delay periods, patient delay, diagnostic delay and treatment delay, in addition to the infectious period after the initiation of treatment (Figure 2).
Defining the onset of TB is crucial when estimating the infectious period. In this study, the date of cough onset is taken as the point of entry into the infectious pool. Cough is the first symptom among approximately 95% of patients with active TB and is the most important variable in TB transmission (Yimer et al. 2005; Lawn et al. 1998; Demissie et al. 2002). We propose that the infectious period is the time during which the patient coughs, and that TB management time defines each patient’s contribution to the infectious pool. Cough is a noticeable symptom, and the time of onset may be recalled without difficulty by most patients. The dates of cough onset and start of treatment are important parameters that define TB management time and can be obtained by simply including an additional question on the unit TB registration forms. The application of the tool can be monitored at the implementing local health facility level by examining the total infectious pool over a specified period of time. A decrease or increase in the infectious pool may indicate changes in the performance of the TB control programme at that level. At present, there is no adequate system for monitoring the infectious pool and TB control programme performance. An alternative method that is simple, inexpensive and can be implemented at the local level is urgently needed to monitor the infectious pool and trends in TB control programme performance. TB management time is a new concept for monitoring the infectious pool of TB in a population. Limitations of the study There are a number of methodological concerns that should be considered. Cough is used as marker of infectiousness in this study. It is expected that some patients may have difficulty in remembering the exact date of cough onset due to longer durations and may therefore be subject to a recall bias. The relative number of undiagnosed cases in different settings may differ. The proportion of the undiagnosed
Patient health seeking and diagnosis at GHF/PHF or others
Cough (TB symptoms)
Patient delay
Diagnosis
Diagnostic delay
Treatment start
Treatment delay
TB management time Figure 2 Relation between different delay periods among patients with TB. GHF: Government health facility; PHF: Private health facility.
318
© 2014 John Wiley & Sons Ltd
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
group used in this article may not be the true figure for our study population for two reasons. Firstly, the local prevalence study that we used to derive our estimate was conducted in 2006 and may not accurately indicate the situation of undiagnosed TB in the study area. Secondly, the size of undetected smear-negative TB has not been estimated; we may thus have underestimated the total size of undiagnosed PTB cases in the study area. This parameter should ideally be measured in a survey of the same population before calculating the infectious pool and full-scale implementation of the tool. The average sputum and culture conversion time used to estimate the infectious period after start of TB treatment was based on previous studies from other countries. This may have resulted in under- or overestimation of the infectious period in our study. More studies addressing the time and factors of smear and culture conversions are needed. The fact that culture was not performed may have resulted in over diagnoses specifically among sputum smear-negative patients, and this may have resulted in overestimation of the size of infectious pool. Conclusions We used a systematic registration of TB management time in unit TB registry books to estimate the infectious pool of TB and determine trends of programme performance at the local level. Additional prospective local studies need to be conducted, and TB management time should be validated as a surrogate marker for programme performance before full-scale implementation of the tool. Potential limitations of this tool must be addressed to secure optimal application. The patient’s ability to recall the duration of symptoms, the size of the undiagnosed group, the contribution of smear-negative cases particularly in high TB burden countries and the contribution of retreatment cases including MDR-TB must be determined to accurately estimate the size of the infectious pool at local level. More research is warranted to develop an acceptable, simple and inexpensive tool to monitor the infectious pool of TB in resource poor settings. Acknowledgements We are grateful to the Research Council of Norway, GLOBVAC Programme (grant 192468/S50), the University of Oslo, the Amhara Regional State Health Bureau and West Gojjam Zone Health Department for funding and facilitating this study. The authors thank Bjørn Haneberg for constructive discussions.
© 2014 John Wiley & Sons Ltd
References Behr MA, Warren SA, Salamon H, et al. (1999) Transmission of Mycobacterium tuberculosis from patients smear-negative for acid-fast bacilli. Lancet 353, 444–449. Brooks SM, Lassiter NL & Young EC (1973) A pilot study concerning the infection risk of sputum positive tuberculosis patients on chemotherapy. The American review of respiratory disease 108, 799–804. CSA (2010) Statistical Tables for the 2007 Population and Housing Census of Ethiopia. CSA, Addis Ababa. Daniel W (1987) Biostatistics: A foundation for analysis in the health sciences, 4th edn. John Wiley & Sons Inc, New York, pp. 155–156. Demissie M, Lindtjørn B & Berhane Y (2002) Patient and health service delay in the diagnosis of pulmonary tuberculosis in Ethiopia. BMC Public Health 2, 23. Dominguez-Castellano A, Muniain MA, Rodriguez-Bano J, et al. (2003) Factors associated with time to sputum smear conversion in active pulmonary tuberculosis. International Journal of Tuberculosis and Lung Disease 7, 432–438. Dye C (2008) Breaking a law: Tuberculosis disobeys Styblo’s rule. Bulletin of the World Health Organization 86, 1. Dye C, Garnett GP, Sleeman K & Williams BG (1998) Prospects for worldwide tuberculosis control under the WHO DOTS strategy. Directly observed short-course therapy. The Lancet 352, 1886–1891. Fitzwater SP, Caviedes L, Gilman RH, et al. (2010) Prolonged infectiousness of tuberculosis patients in a directly observed therapy short-course program with standardized therapy. Clinical Infectious Diseases 51, 371–378. Glaziou P, van der Werf OI & Dye C (2008) Tuberculosis prevalence surveys: rationale and cost. International Journal of Tuberculosis and Lung Disease 12, 1003–1008. Harries AD, Nyirenda TE, Kemp JR, Squire BS, Godfrey-Faussett P & Salaniponi FM, (2003) Management and outcome of tuberculosis patients who fail treatment under routine programme conditions in Malawi. International Journal of Tuberculosis and Lung Disease 11, 1040–1044. Heldal E, Arnadottir T, Cruz JR, Tardencilla A & Chacon L (2001) Low failure rate in standardised retreatment of tuberculosis in Nicaragua: patient category, drug resistance and survival of ‘chronic’ patients. International Journal of Tuberculosis and Lung Disease 5, 129–136. Hernandez-Garduno CV, Kunimoto D, Elwood RK, Black WA & FitzGerald JM (2004) Transmission of tuberculosis from smear negative patients: a molecular epidemiology study. Thorax 59, 286–290. Lawn SD, Afful B & Acheampong JW (1998) Pulmonary tuberculosis: diagnostic delay in Ghanaian adults. International Journal of Tuberculosis and Lung Disease 2, 635–640. Lawn SD, Edwards D & Wood R (2009) Tuberculosis transmission from patients with smear-negative pulmonary tuberculosis in sub-Saharan Africa. Clinical Infectious Diseases 48, 496–497. Mehra RK, Dhingra VK, Nish A & Vashist RP (2008) Study of relapse and failure cases of CAT I retreated with CAT II under
319
Tropical Medicine and International Health
volume 19 no 3 pp 313–320 march 2014
S. A. Yimer et al. TB management in time
RNTCP–an eleven year follow up. The Indian Journal of Tuberculosis 55, 188–191. MOH (2002) National Tuberculosis and Leprosy Prevention and Control Program, Manual. MOH , Addis Ababa, Ethiopia. Ottmani SE, Zignol M, Bencheikh N, La^asri L, Chaouki N & Mahjour J (2006) Results of cohort analysis by category of tuberculosis retreatment cases in Morocco from 1996 to 2003. International Journal of Tuberculosis and Lung Disease 10, 1367–1372. Parikh R, Nataraj G, Kanade S, Khatri V & Mehta P (2012) Time to sputum conversion in smear positive pulmonary TB patients on category I DOTS and factors delaying it. Journal of the Association of Physicians of India 60, 22–26. Rouillon A, Perdrizet S & Parrot R (1976) Transmission of tubercle bacilli: the effects of chemotherapy. Tubercle 57, 275–299. Senkoro M, Mfinanga SG & Mørkve O (2010) Smear microscopy and culture conversion rates among smear positive pulmonary tuberculosis patients by HIV status in Dar es Salaam, Tanzania. BMC infectious diseases 16(10), 210. Shargie EB, Mørkve O & Lindtjørn B (2006) Tuberculosis casefinding through a village outreach programme in a rural setting in southern Ethiopia: community randomized trial. Bulletin of the World Health Organization 84, 112–119. Storla DG, Yimer S & Bjune GA (2010) Can treatment delay be utilized as a key variable for monitoring the pool of infectious tuberculosis in a population? The Journal of Infection in Developing Countries 4, 83–90. Telzak EE, Fazal BA, Pollard CL, Turett GS, Justman JE & Blum S (1997) Factors influencing time to sputum conversion among patients with smear-positive pulmonary tuberculosis. Clinical Infectious Diseases 25, 666–670. Thomas A, Gopi PG, Santha T, et al. (2005) Predictors of relapse among pulmonary tuberculosis patients treated in a
DOTS programme in South India. International Journal of Tuberculosis and Lung Disease 9, 556–561. Tiemersma EW, van der Werf MJ, Borgdorff MW, Williams BG & Nagelkerke NJ (2011) Natural history of tuberculosis: duration and fatality of untreated pulmonary tuberculosis in HIV negative patients: a systematic review. PLoS ONE 6, e17601. Tostmann A, Kik SV, Kalisvaart NA, et al. (2008) Tuberculosis transmission by patients with smear-negative pulmonary tuberculosis in a large cohort in the Netherlands. Clinical Infectious Diseases 47, 1135–1142. UNDP (2012) United Nations, Millennium Development Goals Report. UNDP, New York, pp 44–45. van der Wer MJ, Enarson DA & Borgdorff MW (2008) How to identify tuberculosis cases in a prevalence survey. International Journal of Tuberculosis and Lung Disease 12, 1255–1260. WHO (2006) The Stop TB strategy. WHO/HTM/TB/2006.368, WHO, Geneva. WHO (2009) Management of tuberculosis training for health facility staff, 2nd edn. WHO/HTM/TB/2009.423c, WHO, Geneva. WHO (2011) Global tuberculosis control: WHO Annual Report (WHO/HTM/TB/2012.6), WHO, Geneva. Yimer S, Bjune G & Alene G (2005) Diagnostic and treatment delay among pulmonary tuberculosis patients in Ethiopia A cross-sectional study. BMC Infectious Diseases, 5, 112. Yoshiyama T, Shrestha B & Maharjan B (2010) Risk of relapse and failure after retreatment with the Category II regimen in Nepal. International Journal of Tuberculosis and Lung Disease 14, 1418–1423. Zeleke A & Fasil DAK (2011) Ethiopian national tuberculosis prevalence survey 2010–2011: Preliminary Result: Paper presented at the 42nd Union World Conference on Lung Health; 2011 Oct 26–30; Lille, France.
Corresponding Author Solomon A Yimer, Division of Infectious Disease Control, Department of Bacteriology and Immunology, Norwegian Institute of Public Health, PO Box 4404 Nydalen, 0403 Oslo, Norway. Tel.: +47 47687670; +47 21077000; Fax: +47 22353605; E-mail: [email protected]
320
© 2014 John Wiley & Sons Ltd
