64th ISI World Statistics Congress - Ottawa, Canada

64th ISI World Statistics Congress - Ottawa, Canada

Predicting the pulmonary tuberculosis treatment outcome using the Scoring method

Author

AN
Dr Andre Nana-Yakam

Co-author

  • AN
    Dr Andre Nana-Yakam
  • J
    Joseph Magloire Fossokeng Mouafo
  • C
    Claude Simo
  • J
    Jules Sadefo Kamdem
  • S
    Samuel Bowong
  • L
    Louis Aimé Fono
  • J
    Jürgen Noeske

Conference

64th ISI World Statistics Congress - Ottawa, Canada

Format: CPS Poster

Keywords: agenda2030, communicable, disease, health, illnesses,, modelling, multivariate, score

Abstract

Background: Tuberculosis is the largest killer among communicable diseases in the 15 to 49 years age group, when humans are most productive. In 2021, there were an estimated 10 million new TB cases worldwide. One of the targets of the Sustainable Development Goals (SDGs) for 2030, is to end the global tuberculosis (TB) epidemic. Failure to properly treat drug susceptible tuberculosis (DSTB) can leads to the acquisition of multi drug resistance TB (MDRTB). The cost per patient treated is usually in the range of US$100 to 1000 for DSTB and US$2000 to 20000 for MDRTB. Meaning for example that, in addition to the US$2.66 to 26.57 million needed for the annual treatment of DSTB patients in Cameroon, it will cost US$0.75 to 7.44 million to treat the estimated 1.4% MDRTB patients that could result from the poor outcome treatment. Preventing treatment failures of DSTB is therefore both a health and financial issue. The decision support tool, which can measure the risk of unsuccessful treatment outcome, very popular and worldwide used in the domain of finance, is not yet available or popularized to prevent a favourite treatment outcome of infectious diseases such as tuberculosis. The objectives of this study were to use the Statistical method of Scoring to identify factors associated with the tuberculosis unsuccessful treatment outcome and to predict the treatment outcome.

Methods:
The data used in this study was provided by a survey undertaken between 1st of May 2011 and 30th of April 2012 in the city of Douala, Cameroon. The study population was the smear positive pulmonary tuberculosis patients. At enrolment, all study participants answered a structured questionnaire. Patients enrolled were followed up until the end of their treatment. The outcomes of their treatment were registered and classified according to WHO definitions. A sample of 1,529 patients were randomly selected among the 2,132 enrolled patients. This sample was randomly split into two parts: one subsample of 1,200 patients (78%) used as the Development sample, and the remaining of 329 patients (22%) used as the Validation sample. Baseline characteristics associated with unsuccessful treatment outcomes were investigated using logistic regressions. The optimal score was numerically computed. A model used to estimate the risk of unsuccessful treatment outcome was derived. The threshold probability which maximize the area under the ROC curve was 18%. Using the validation sample, patients for whom the risk was greater than this threshold were classified as unsuccessful treatment outcome and the others as successful.

Results: HIV positive status, active smoker and non-belief in TB healing were the factors significantly associated with unsuccessful treatment outcomes (p < 0.05). HIV positive and active smoking status were associated with death; the non-belief in healing, being youth and male gender were associated with lost-to-follow-up, having a TB antecedent and not having TB contact were associated with therapeutic treatment failure.

Conclusion: To increase the tuberculosis treatment success' rate, additional measures should be taken during the follow-up and the treatment of targeted TB patients presenting high identify unsuccessful treatment outcomes risk characteristics.

Figures/Tables

Table 1

table2

Table 3

roc05