Mathematical modelling to project the impact of interventions targeted to previously treated individuals on the trajectory of the tuberculosis epidemic in high tuberculosis prevalence settings
| dc.contributor.advisor | Beyers, Nulda | en_ZA |
| dc.contributor.advisor | Cohen, Theodore | en_ZA |
| dc.contributor.author | Marx, Florian Michael | en_ZA |
| dc.contributor.other | Stellenbosch University. Faculty of Medicine and Health Sciences. Paediatrics and Child Health. | en_ZA |
| dc.date.accessioned | 2018-02-27T12:13:51Z | |
| dc.date.accessioned | 2018-04-09T11:48:12Z | |
| dc.date.available | 2018-12-31T03:00:12Z | |
| dc.date.issued | 2018-03 | |
| dc.description | Thesis (PhD)--Stellenbosch University, 2018. | en_ZA |
| dc.description.abstract | ENGLISH ABSTRACT: Better strategies are needed to reduce Mycobacterium tuberculosis (M.tb) transmission in tuberculosis (TB) high-incidence settings. Targeting additional control interventions towards groups at highest TB risk may thereby constitute a reasonable approach to make best use of limited resources. One group considered at high TB risk are people with a history of previous TB treatment. In recent years, high rates of recurrent TB after successful TB treatment, frequently due to reinfection with another M.tb strain, have been reported from several highincidence settings. The extent to which previously treated people contribute to the overall TB burden in these settings, and whether control interventions targeted towards this high-risk group could help reduce transmission has not been established. The aim of the research presented in this dissertation was to characterise the risk of TB among people with a history of previous TB treatment, and to project the impact of control interventions targeted at this subgroup on the trajectory of TB epidemics in high-incidence settings. Towards this aim, I conducted a total of five studies, four of which are based on traditional epidemiological analysis (Chapters 2,3, and 5,6), and one used a transmission-dynamic mathematical model (Chapter 4). Two studies were based on routine TB program data from a high-incidence setting in suburban Cape Town. The first aimed at estimating the rate of re-treatment for smear-positive TB dependent on whether individuals had completed their previous TB treatment (Chapter 2). For the second, I used M.tb strain-type DNA fingerprinting data for a subset of patients re-treated for recurrent smear-positive TB to investigate the relationship between the type of recurrence, i.e. endogenous reactivation (relapse) or exogenous reinfection, and the time to recurrent smear-positive TB (Chapter 3). I found, among 2,136 smear-positive TB patients, a high rate of re-treatment for smear-positive TB among those who had been lost to follow-up from treatment (6.86 [5.59-8.41] per 100 PYRS). By the end of the second year 28% had been diagnosed again with smear-positive TB. The rate was 3-5-times higher (aHR: 3.97 [3.00 - 5.26]) than after treatment success (cure: 2.09 [1.81-2.41]). Among those who had successfully completed TB treatment, the rate of smear-positive TB was at least 4-times higher than the estimated rate of new smear-positive TB in this setting. Individuals after successful treatment accounted for the majority (68%) of TB patients re-treated for smear-positive TB (Chapter 2). Exogenous reinfection was the underlying mechanism of disease recurrence in 66 (51%) of 130 individuals re-treated for smear-positive TB after previous treatment success. The proportion did not change after restricting the analysis to individuals with documented HIVnegative test results (27 [51%] reinfections of 53 recurrences). The rate of relapse was highest 4-5 months after treatment completion, whereas the rate of reinfection TB dominated after the first year and remained high for several years (Chapter 3). I then used a calibrated transmission-dynamic mathematical model to project the populationlevel impact of interventions targeted at previously treated people in the same high-incidence setting. The interventions modelled were annual targeted active case finding (TACF) and secondary isoniazid preventive therapy (2°IPT) among people who had successfully completed TB treatment. The model projected that, under current control efforts, local TB incidence will remain in slow decline for at least another decade in this setting, and that interventions targeted at previously treated people would greatly accelerate this decline. Annual TACF combined with 2°IPT was projected to avert 40% (20%-59%) of incident TB cases and 41% (16%-62%) of TB deaths estimated to occur in the local population between 2016 and 2025 (Chapter 4). While the previous 3 studies (Chapters 2-4) had focussed on a particular TB high-incidence setting in suburban Cape Town, I conducted two further studies to explore whether specific characteristics of previously treated TB observed in suburban Cape Town extend to other highincidence settings in Southern Africa. I analysed TB prevalence survey data for more than 64,000 adults in 8 South African and 16 Zambian communities, to estimate TB prevalence stratified by history of previous TB treatment, and to investigate the extent to which previously treated people contributed to the overall prevalent TB burden. The study revealed a high prevalence of bacteriologically-confirmed TB among previously treated people in the 8 South African (overall: 3.81% [95%CI 3.25%–4.47%]) and the 16 Zambian (1.01% [95%CI: 0.65%–1.55%]) communities. Previously treated people accounted for 20.7% and 10.4% of prevalent TB cases in the South African and Zambian communities, respectively, and for more than 20% in 9 of the 24 communities overall (Chapter 5). Finally, I made use of electronic TB register data from the 52 South African health districts to investigate the proportion of previously treated individuals among notified TB patients treated for bacteriologically-confirmed TB in 2011. The study showed that the proportion of previously treated TB varied in the 52 health districts between 7.6% and 40%. The proportion exceeded 20% in 17 of the 52 districts. Higher proportions of previously treated TB correlated with higher TB case notification rates (r=0.75; P<0.001) and lower estimates of HIV prevalence (r=-0.45; P<0.001) in the districts (Chapter 6). In conclusion, this research documents high rates of TB after previous loss to follow-up from treatment and after treatment success in a high-incidence setting. People who previously completed TB treatment constitute the majority of smear-positive TB re-treatment patients, suggesting that efforts to ensure treatment adherence alone are unlikely to be sufficient to reduce the TB burden among previously treated people. Our study is consistent with earlier findings from this setting that reinfection contributes considerably to recurrent TB, even among HIV-uninfected individuals. I document for the first-time distinct temporal dynamics of relapse and reinfection TB. These dynamics suggest that sampling bias/differences in follow-up time is likely to explain the substantial variation in the contribution of reinfection to disease recurrence reported in observational studies. High rates of reinfection TB over a lengthy time after treatment completion suggest that the performance of TB treatment alone is unlikely to explain the high burden of recurrent TB in this setting. I used a transmission-dynamic mathematical model to address the idea of a targeted control strategy at a time when novel strategies are urgently needed to reduce transmission and the TB burden in populations most severely affected by TB. The model suggests considerable public health potential for TB control interventions targeted at previously treated people in this high-incidence setting. I identified several other high-incidence communities (and districts) where previously treated individuals contribute considerably to the prevalent and incident (notified) TB burden, and where interventions in this high-risk group might be relevant. This work represents only a first step towards evaluating the potential of this targeted control approach. Further research will be necessary to determine the feasibility, impact and costeffectiveness of targeting interventions towards previously treated people for reducing transmission and the TB burden in high-incidence settings. | en_ZA |
| dc.description.abstract | AFRIKAANSE OPSOMMING: Beter strategieë word vereis om die oordrag van Mycobacterium tuberculosis (M.tb) in omgewings met ’n hoë tuberkulose (TB) insidensie te verminder. Die toespitsing van bykomende beheerintervensies op groepe met die hoogste TB-risiko kan dus’n redelike benadering wees om beperkte hulpbronne so doeltreffend moontlik te benut. Een groep wat as hoë TB-risiko beskou word, is persone met ’n geskiedenis van vorige TB-behandeling. Die afgelope paar jaar is ’n groot aantal gevalle van wederkerende TB ná suksesvolle TBbehandeling, dikwels as gevolg van herinfeksie met ʼn ander M.tb-stam, in verskeie hoëinsidensieomgewings aangemeld. Die mate waarin voorheen behandelde persone tot die algehele TB-las in hierdie omgewings bydra, en of beheerintervensies met die oog op hierdie hoë-risikogroep oordrag kan help verminder, is nog nie vasgestel nie. Die doel van die navorsing wat in hierdie proefskrif aangebied word, was om die TB-risiko onder persone met ’n geskiedenis van vorige TB-behandeling te beskryf, en om te voorspel watter impak beheerintervensies onder hierdie subgroep op die trajek van TB-epidemies in hoë-insidensieomgewings sal hê. Vir dié doel het ek altesaam vyf studies onderneem, waarvan vier op tradisionele epidemiologiese ontleding (hoofstuk 2, 3, 5 en 6) en een op ’n dinamiese wiskundige oordragmodel berus (hoofstuk 4). Twee studies was gegrond op roetine TB-programdata uit ’n hoë-insidensieomgewing in voorstedelike Kaapstad. Die doel van die eerste studie was om die herbehandelingskoers vir smeerpositiewe TB te bepaal na gelang van of individue hulle vorige TB-behandeling voltooi het (hoofstuk 2). Vir die tweede studie het ek M.tb-stamtipe DNS-vingerafdrukdata gebruik van ’n subgroep pasiënte wat opnuut vir wederkerende smeerpositiewe TB behandel is om die verwantskap tussen die tipe wederkerende TB – d.w.s. endogene heraktivering (relaps) of eksogene herinfeksie – en die tydverloop tot en met wederkerende smeerpositiewe TB te bepaal (hoofstuk 3). Ek het bevind dat uit ’n groep van 2136 smeerpositiewe TB-pasiënte was die herbehandelingskoers vir smeerpositiewe TB hoog onder diegene wat hulle opvolg van behandeling nagelaat het (6,86 [5,59-8,41] per 100 PJ). Teen die einde van die tweede jaar was 28% reeds weer met smeerpositiewe TB gediagnoseer. Die koers was drie tot vyf keer hoër (aHR: 3,97 [3,00-5,26]) as ná behandelingsukses (genees: 2,09 [1,81-2,41]). Onder diegene wat hulle TB-behandeling suksesvol voltooi het, was die smeerpositiewe TB-syfer ten minste vier keer hoër as die geskatte nuwe smeerpositiewe TB-syfer in hierdie omgewing. Individue wat voorheen suksesvol behandel is, het die meerderheid (68%) uitgemaak van TBpasiënte wat weer vir smeerpositiewe TB behandel is (hoofstuk 2). Eksogene herinfeksie was die onderliggende meganisme van wederkerende siekte by 66 (51%) uit 130 individue wat ná vorige suksesvolle behandeling weer vir smeerpositiewe TB behandel is. Die gedeelte het onveranderd gebly selfs nadat die ontleding beperk is tot individue met gedokumenteerde MIV-negatiewe toetsresultate (27 [51%] herinfeksies uit 53 gevalle van wederkerende siekte). Die relapssyfer was die hoogste sowat vier tot vyf maande na die voltooiing van behandeling, terwyl die TB-herinfeksiesyfer ’n hoogtepunt bereik het na die eerste jaar, en vir ’n hele aantal jaar daarna hoog gebly het (hoofstuk 3). Daarna het ek ’n gekalibreerde dinamiese wiskundige oordragmodel gebruik om die populasievlakimpak van intervensies vir voorheen behandelde persone in dieselfde hoëinsidensieomgewing te voorspel. Die gemodelleerde intervensies was jaarlikse toegespitste aktiewe gevalleopsporing (“TACF”) en sekondêre isoniasied-voorkomingsbehandeling (“2°IPT”) onder persone wat TB-behandeling suksesvol voltooi het. Die model dui daarop dat, met huidige beheerpogings, plaaslike TB-insidensie in hierdie omgewing vir minstens nog ’n dekade stadig sal daal, en dat intervensies met die oog op voorheen behandelde persone dié daling beduidend sal versnel. Volgens die model sal jaarlikse TACF in samehang met 2°IPT sowat 44% (20-59%) van nuwe TB-gevalle en 41% (16-62%) van TB-sterftes verhoed wat na raming tussen 2016 en 2025 in die plaaslike populasie sal voorkom (hoofstuk 4). Terwyl die drie studies hierbo (hoofstuk 2-4) op ’n bepaalde voorstedelike Kaapse omgewing met ’n hoë TB-insidensie gekonsentreer het, het ek ook twee verdere studies gedoen om te bepaal of sekere kenmerke van voorheen behandelde TB wat in voorstedelike Kaapstad opgemerk word, ook in ander hoë-insidensieomgewings in Suider-Afrika voorkom. Hiervoor het ek TB-prevalensieopnamedata van meer as 64000 volwassenes in agt Suid- Afrikaanse en 16 Zambiese gemeenskappe ontleed om TB-prevalensie na gelang van vorige TB-behandelingsgeskiedenis te beraam, en om vas te stel in watter mate voorheen behandelde persone tot algehele TB-prevalensie bydra. Die studie dui op ’n hoë prevalensie van bakteriologies bevestigde TB onder voorheen behandelde persone in die agt Suid- Afrikaanse (algeheel: 3,81% [95%CI 3,25-4,47%]) sowel as die 16 Zambiese gemeenskappe (1,01% [95%CI: 0,65-1,55%]). Voorheen behandelde persone maak onderskeidelik 20,7% en 10,4% van TB-prevalensie in die Suid-Afrikaanse en Zambiese gemeenskappe uit, en meer as 20% in nege van die 24 gemeenskappe altesaam (hoofstuk 5). Laastens het ek elektroniese TB-registerdata van die 52 Suid-Afrikaanse gesondheidsdistrikte gebruik om die persentasie voorheen behandelde individue te bepaal onder aangemelde TBpasiënte wat in 2011 vir bakteriologies bevestigde TB behandel is. Hierdie studie toon dat die persentasie voorheen behandelde TB in die 52 gesondheidsdistrikte tussen 7,6% en 40% wissel. In 17 van die 52 distrikte oorskry die persentasie 20%. Hoër persentasies voorheen behandelde TB korreleer met hoër TB-aanmeldingsyfers (r = 0,75; P<0,001) en laer MIVprevalensieskattings (r = -0,45; P<0,001) in die distrikte (hoofstuk 6). Ten slotte bevind hierdie navorsing hoë TB-syfers ná vorige verlies van opvolg tydens behandeling sowel as ná suksesvolle behandeling in ’n hoë-insidensieomgewing. Persone wat voorheen TB-behandeling voltooi het, maak die meerderheid uit van smeerpositiewe TBherbehandelingspasiënte wat daarop dui dat pogings om slegs behandelingsgetrouheid te verseker waarskynlik nie voldoende is om die TB-las onder voorheen behandelde persone te verminder nie. Ons studieresultate strook met vroeëre bevindinge in hierdie omgewing dat herinfeksie beduidend tot wederkerende TB bydra, selfs onder MIV-negatiewe individue. Ek dokumenteer ook vir die eerste keer die tyddimensies van TB-relaps en -herinfeksie. Hierdie dimensies dui daarop dat sydigheid in steekproefneming/tydverskille in die duur van opvolg waarskynlik die aansienlike variasie in die bydrae van herinfeksie tot wederkerende siekte in waarnemingstudies verklaar. Hoë TB-herinfeksiesyfers oor ’n lang tydperk ná voltooiing van behandeling dui daarop dat TB-behandelingsprestasie alleen waarskynlik nie die hoë las van wederkerende TB in hierdie omgewing verklaar nie. My gebruik van ’n dinamiese wiskundige oordragmodel om die gedagte van ’n toegespitste beheerstrategie te ondersoek, val saam met ’n dringende behoefte aan nuwe strategieë om TB-oordrag en die TB-las te verminder by populasies wat die ergste deur dié siekte geraak word. Die model dui op aansienlike openbaregesondheidspotensiaal vir TBbeheerintervensies wat op voorheen behandelde persone in hierdie hoë-insidensieomgewing gerig is. Ek identifiseer ook verskeie ander hoë-insidensiegemeenskappe (en -distrikte) waar voorheen behandelde individue beduidend tot TB-prevalensie en aangemelde TB-insidensie bydra, en waar intervensies vir hierdie hoë-risikogroep moontlik relevant kan wees. Tog is hierdie werk maar die eerste stap om die potensiaal van hierdie toegespitste beheerbenadering te beoordeel. Verdere navorsing word vereis om die haalbaarheid, impak en kostedoeltreffendheid van toegespitste intervensies vir voorheen behandelde persone te bepaal in die strewe na laer TB-oordrag en ’n kleiner TB-las in hoë-insidensieomgewings. | af_ZA |
| dc.embargo.terms | 2018-12-31 | |
| dc.format.extent | 120 pages | en_ZA |
| dc.identifier.uri | http://hdl.handle.net/10019.1/103926 | |
| dc.language.iso | en_ZA | en_ZA |
| dc.publisher | Stellenbosch : Stellenbosch University | en_ZA |
| dc.rights.holder | Stellenbosch University | en_ZA |
| dc.subject | Mathematical modelling | en_ZA |
| dc.subject | Tuberculosis -- Diagnosis | en_ZA |
| dc.subject | Tuberculosis in children | en_ZA |
| dc.subject | Tuberculosis -- Prevention | en_ZA |
| dc.title | Mathematical modelling to project the impact of interventions targeted to previously treated individuals on the trajectory of the tuberculosis epidemic in high tuberculosis prevalence settings | en_ZA |
| dc.type | Thesis | en_ZA |