Screening and prevention of Tuberculosis : the impact of screening policy and practice on public health outcomes

Abstract

Tuberculosis (TB) is a disease of public health importance due to the associated morbidity and mortality as well as the potential for transmission of infection to uninfected contacts. Tuberculosis can be prevented by screening of high risk groups and targeted prophylaxis with isoniazid (INH). Screening programs with structured guidelines exist for the screening of many groups at risk for TB, including refugees, contacts, health workers and prisoners, with some run by health authorities and others by employers. Screening programs for TB may focus on case finding, prevention or a combination of both, and may differ in their aims, guidelines, practice and resources. The outcomes and effectiveness of such programs may also differ, and may be influenced by a number of factors. These include the nature of screening guidelines and whether such guidelines are evidence based, the consistency between policy and practice of screening by health professionals, and external factors such as structural or political change and financial restrictions. The null hypothesis is that TB screening programs operate under evidence based guidelines, have good implementation of guidelines and have optimal outcomes. Aims To examine screening programs for contacts, refugees and prisoners in selected programs in Australia and the USA. Because of the differences in screening methods and policy outlined above, I aimed to examine the rationale of screening policies and their scientific basis, the efficacy of screening practice, whether screening guidelines are properly implemented by medical and nursing staff, what intrinsic and extrinsic problems may have hindered the implementation of guidelines, and the outcomes of screening programs as measured by the incidence of active TB and skin test conversion, and the prevention of active TB. Methods I undertook a series of separate studies in order to address the aims above. The studies are outlined below. The first chapter presents background and a general literature review, but more specific literature reviews are contained in the subsequent chapters. 1) The first study retrospectively estimates missed opportunities for prevention of TB in notified cases of active TB in Victoria, Australia. All notified cases of active TB in 1991 in Victoria (n=231) were reviewed for any past screening for TB. If such screening or indications for screening were documented, each case was evaluated as to whether appropriate preventive action was taken. Missed opportunities for prevention of TB were quantified for each case. 2) The next chapter is a retrospective cohort study of 1,142 contacts of tuberculosis screened in 1991. This cohort was screened by health authorities in Victoria, Australia, and had two years of follow up for the development of active disease. The study describes the two-year incidence of TB in recent contacts. Skin test reaction sizes were correlated with risk of developing active disease in order to provide data to select appropriate criteria for the interpretation of the test. Preventability of incident cases of TB was quantified retrospectively. 3) The third study is a detailed analysis of each step of screening and prevention in the 1,142 recent contacts described in item (2) above. This study compares recommended guidelines with actual practice. 4) The cost effectiveness of contact screening per case prevented and per case found was estimated for three different models. The first model describes cost effectiveness of contact screening as it was conducted in 1991; the second model describes cost effectiveness of contact screening had the 1991 guidelines been followed closely; and the final model describes the cost effectiveness of contact screening practised according to hypothetical, evidence-based guidelines. 5) This study is a retrospective cohort study of 1,101 Indo-Chinese refugees screened between 1989-1990 in Victoria, Australia. The study included five years of follow up for the development of active disease. Skin test reaction sizes were correlated with risk of developing active disease in order to provide data to select appropriate criteria for the interpretation of the test. Preventability of incident cases was assessed retrospectively. 6) This study is a detailed analysis of each step of screening and prevention in the 1,1 01 refugees, comparing recommended guidelines with actual practice. 7) This study is a comparison of tuberculin skin test distributions in a number of Victorian populations of varying degree of risk for TB. The skin test distributions are used to assess the sensitivity and specificity of the cut-off points used in Victoria at the time for defining a positive reaction. 8) This is a study to determine the incidence of skin test conversion during annual screening of inmates of Maryland prisons, with an analysis of the impact of TB control measures and risk factors on the incidence of skin test conversion. 9) A study of the skin test converters identified in (8) above was then conducted to determine exposure to undetected TB within the correctional system. The movements within the correctional system of skin test converters were matched with those of inmates with documented infectious TB. This was supplemented by a linkage study with the State TB registry to identify cases of active TB which may have been missed within the prison system. 1 0) In the same group, I studied implementation of screening guidelines by prison medical and nursing staff and the utilisation of INH prophylaxis. Results The study of all notified cases of active TB in 1991 identified that nearly half of all notified cases of TB have been screened for TB in the past, and yet this screening had failed to prevent the development of active TB. Over 70% of those screened were found to be at risk for TB, but rates of preventive therapy were low. Nearly 30% of cases may have been prevented. There was evidence that the guidelines for screening and prevention used in Victoria for refugees and contacts required updating. These guidelines emphasise case finding rather than identification of asymptomatic infection and prevention, and rely more on chest radiograph (CXR) screening than skin testing. In addition, the guidelines were poorly implemented . This was partly explained by devolution of TB screening programs in the 1970s and 1980s and lack of resources. In nearly 60% of contacts, the presence or absence of infection could not be determined because a skin test was not done, and a CXR, if done, was clear. The rate of preventive therapy for eligible contacts and refugees was low. The two year incidence of active pulmonary TB was 5311100,000 per year for contacts, and 110/100,000 per year for refugees. Of the incident cases, many were considered "not infected" at the time of screening because of they had received BCG vaccine in the past and had a skin test reaction of 1 0-19mm. In logistic regression models testing various skin test cut-off points, a reading of 15mm or more was the strongest predictor of the development of active TB for contacts. The use of a 20mm cut-off excludes most individuals who are at risk. The incidence of active TB increased with increasing skin test reaction size. The direct cost of contact screening as it was actually performed in 1991 was A$331,610 per case prevented, A$248,708 per case found and A$871 per contact traced. If the guidelines had been followed closely, the respective costs would have been A$81 ,892 per case prevented, A$286,621 per case found and A$1 ,004 per contact traced. In an alternative model which I propose, the costs would be A$47,358 per case prevented, A$319 ,670 per case found and A$793 per contact traced. The guidelines used in Maryland prisons, in contrast, are well supported by the literature. However, I found a poor implementation of guidelines by prison medical and nursing staff, and a low rate of preventive therapy for eligible inmates. Skin test conversions occurred at a rate of 6.3/100 person-years, and the rate of preventive therapy was lowest for this high risk group. Reasons for not giving, or prematurely ceasing INH were largely unfounded. The rate of true side effects was low. There was a wide variation in skin test conversion rates and in rates of giving INH, between different prisons and prison types. The highest risk of conversion was in the intake institution. A strong positive correlation existed between prison crowding and skin test conversion, and a strong negative correlation with rates of INH use and skin test conversion. Exposure to TB was only found for 30% of skin test converters. Linkage with the State TB registry identified cases of TB that occurred in inmates but were undiagnosed during incarceration. Discussion The lack of adherence to screening guidelines and the inadequate use of preventive therapy were problems common to all screening programs. Although under-use of preventive therapy in different settings has been reported in the literature, this is a poorly studied area. The low rates of preventive therapy and the low threshold for discontinuing it once started, suggest that confusion and fear about the use of INH may be prevalent, and that there is a need for education of providers. The guidelines used for contact and refugee screening in Victoria are not well supported by scientific rationale and exclude a large proportion of individuals who are at high risk. This was confirmed by the high incidence of active TB at follow up, and by examining the features of each incident case. It was further confirmed by finding the high rate of preventability of notified cases of active TB. I found a lack of cost effectiveness of the contact screening program, largely because intervention in the form of prevention was an unlikely outcome at the end of a sequence of screening tests. The findings also emphasise that in a low prevalence setting, case finding is an expensive exercise and should not be the main focus of screening. In the Maryland prisons, whilst the guidelines used were well founded, a wide variation was found in adherence to guidelines and in rates of preventive therapy. Evidence was also found that significant transmission ofTB may go undetected in prisons, due to high population turnover. For the Victorian contact and refugee programs, the recommendation that more sensitive skin test criteria be adopted, that age should not be considered for contacts of TB when skin testing or offering preventive therapy, that CXR screening needs to be rationalised, and that contacts of non-infectious TB need not be screened en masse, have been accepted and adopted. These data contributed to significant revision of the screening guidelines in 1994. It would be prudent to conduct a follow up study of the application and impact of these changes. In Maryland prisons, it was recommended that efforts be made to standardise preventive efforts across different prisons, and to prioritize TB control measures according to the level of risk of the prison. Crowding is not a risk factor which is readily amenable to change, but maximal use of preventive therapy can diminish that risk. Guidelines used by screening programs are variable. In the case of Victoria, there was no convincing evidence or data for the use of their screening guidelines. This study provided sound data for decision making and had a favourable impact on the revision of guidelines. This in itself is a public health exercise which proves that long standing policies should be questioned for validity, and if those policies are not supported by evidence, the collection of required evidence for change is indicated. The major problem faced by the screening programs studied, however, is not a lack of policies, but a failure to apply structured guidelines in the practice of prevention. In the case of Victoria, this is partly explained by factors extrinsic to the TB program, such the devolution of TB services in the 1970s and 1980s, and lack of adequate resources and support for TB program staff. There is, nonetheless, considerable opportunity to improve the outcome of screening, which should not be carried out unless there is a commitment to intervention that makes a positive public health impact. Screening without an end point of change in outcome is not good practice. In the case of TB, intervention in the form of INH preventive therapy is available, and as such, should be used.