The community-based ACF approach employed in Cambodia, which focussed on TB contacts in poor communities with a high burden of TB, was found to be cost-effective, considering that the overall cost per case detected was US$ 108 and the NNS was 38. In comparison, eight FIDELIS projects (Fund for Innovative DOTS Expansion through Local Initiatives to Stop TB), supported by the Canadian International Development Agency (CIDA), reported a cost per case detected ranging from US$ 60 to US$ 1626 by ACF [13
], which implies that the Cambodian ACF approach is comparably cost-effective. Besides, the NNS of 38 is very similar to those reported through intensified case finding in HIV-prevalent and congregate settings [14
]. Considering the low HIV prevalence in this study population, the low NNS further justifies the cost-effectiveness of the strategy. The results of this study also revealed that the ACF approach could contribute to the detection of smear-negative TB cases or smear-positive TB cases with low smear grades. The approach could also detect more TB patients from older age groups, which demonstrates an “equalizing role” for ACF in TB service provision, especially as the elderly run the risk of dying without any attempt to diagnose the underlying source of their complaints or disease.
The elderly are generally considered vulnerable to TB as they belong to age cohorts with high TB infection rates, have relatively low immune status and limited access to health services [15
]. A survey conducted in Cambodia in 2002 demonstrated that TB prevalence was up to seven times higher in the older age groups than in the general population. Yet the level of case detection among them was low [17
]. In that sense, detecting TB among older patients could be seen as an additional value of ACF in the Cambodian context. Studies conducted in South India and Eastern Nepal showed similar results [9
]. Some may argue that ACF in Cambodia missed out part of the younger population since most of the sessions were conducted during the day, which is more convenient for the elderly and not for the economically active age group. Yet one of the primary objectives of ACF is to identify and treat more patients who have remained undetected in the community with special attention to health inequity. Given that more than half of actively detected patients were older than 55 years of age compared to around one third of passively detected patients being older than 55 years of age, it is reasonable to conclude that the ACF approach has successfully complemented routine PCF in Cambodia by detecting older patients. The finding that all of the actively detected patients were new cases without any history of previous treatment further supports its complementary function.
Smear grade, a quantitative measure of TB bacilli in the sputum samples, has been suggested as a parameter that can represent the severity and the infectiousness of the disease [19
]. The lower smear grades observed in the ACF group in this study, therefore, suggest that actively detected patients were at a relatively early stage of their disease. This was consistent with the findings of Santha et al. in South India [9
]. Another study exploring the association between smear grade and diagnostic delay in Japan showed that patients with delayed diagnosis had significantly higher smear grades than those with timely diagnosis [21
]. This finding implies that there is a link between smear positivity and timing of diagnosis, and further supports our argument that ACF can contribute to early case finding in a developing country setting. Looking at the newly detected cases among contacts of index smear-positive cases, Liippo et al. concluded that limiting contact investigation to close contacts of patients heavily positive by sputum smear makes contact tracing more effective [19
]. From a different angle, Lin et al. examined the effect of treatment delay on secondary TB infection among household contacts [22
]. The study demonstrated a linear positive relationship between the delays in treatment of index cases and the secondary attack rates among household contacts. Thus, both existing literature and our findings strongly suggest that ACF can reduce diagnostic delay and thereby reduce the risk of further transmission.
As programmatic operational research using routine programme data, our study has several limitations. It is important to note that smear grade is both directly and indirectly influenced by other factors including volume and quality of sputum specimens, capacity of laboratory technicians, and possibly characteristics of screened subjects such as HIV and smoking status [23
]. However, as smear slides were read by local technicians under routine conditions within the standard external quality assurance system, these factors may not have been serious confounders. Nevertheless, it may be desirable to obtain additional evidence on ACF by conducting, for instance, a delay study (comparing the extent of various delays from onset of symptoms, treatment seeking, final diagnosis, to treatment initiation between ACF and PCF) to reconfirm our findings and further quantify its contribution to early case detection.
Another limitation may have been the diagnosis of smear-negative TB during the outreach sessions. Our findings, showing a larger proportion of smear-negative TB detected in ACF than in PCF, might have been partly due to the different diagnostic methods used in between ACF and PCF, and over-diagnosis of smear-negative TB during the ACF sessions. A similarly high proportion of smear-negative TB was reported in a study conducted in South India [9
]. To overcome this limitation, CENAT has been implementing a new project employing Xpert MTB/RIF as more sensitive diagnostic tool.
Our data lack the information on the number of households visited and individuals interviewed in the communities which could serve as primary denominators of NNS. However, our results are still valuable by using the number of individuals attended in the ACF sessions as shown in the result.
Finally, this study was unable to perform a comparative analysis between ACF and PCF in terms of cost-effectiveness since cost calculation of PCF requires an extensive assessment of the health system costs. A proper cost-effectiveness analysis should also include the assessment of disability-adjusted life-years (DALY) lost under the conditions with or without ACF. Obviously, these are beyond the scope of this field-based operational research study.
As one of the negative aspects of ACF, the issue of initial defaulters has been discussed in several articles [4
]. Refusal to start treatment might be attributed to a lack of motivation due to the absence or mildness of symptoms [9
]. Although the available evidence is limited and its definition may vary depending on researchers, the initial defaulter rates in ACF reported in other countries ranged from 26% to 32% [9
], which was much higher than our observation of 5.2%. Taking into account that the TB programme in Cambodia has maintained a high treatment success rate and a low defaulter rate at national level over the years [2
], these programmatic strengths may be reflected in the low initial defaulter rate among actively detected cases. This assumption suggests that ACF is likely to result in more initial defaulters in places with poor TB programmatic indicators, and further suggests that, when adopting ACF, such indicators need consideration in order to maximize effectiveness of the activities.
In this study, the smear grades of patients diagnosed through PCF after the ACF sessions were as low as those of actively detected patients, which might imply that ACF had a sustained impact on smear positivity among patients in the community. The one-time ACF sessions identified and treated many patients with relatively mild symptoms as well as those with heavy bacterial load. This may have contributed to the reduction of the overall patient pool in the community and thus led to the lower smear grades among patients in routine PCF. In addition, taking into account that the previous national prevalence survey was conducted in 2002, those reviewed in the study had not been exposed to massive ACF or TB awareness campaigns for a substantial period of time. Thus our ACF activities were likely to have contributed to increased community awareness about the disease and availability of services. Community mobilization might have promoted early health seeking and diagnosis.