In our 2 models comparing 4HR to 6HE, 4HR is associated with a lower expected mortality rate as well as lower total cost. The first model does not consider the cost of MDR-TB cases, but rather just the cost of the TB treatment regimens in new patients and in those requiring retreatment. In estimates from the WHO, a decreased cost of $4–10 was estimated 
. Our estimate of $2.42 per patient falls slightly below this range. More importantly, because the rate of relapse or failure is higher in patients who take 6HE based on data from clinical trials, this increased likelihood of an unfavorable outcome may lead to an increased risk of MDR-TB. When we considered the cost of MDR-TB in our second model, the cost difference was even larger ($12.74 per patient). Thus, using the WHO estimate of 40,000 new TB cases annually in Uganda, our model results suggest that adopting 4HR may reduce the annual TB-related mortality burden by ~1,800 patients and may lead to expected savings of approximately $100,000–$500,000 per year to the Ugandan national health care system.
Our estimate of the number of cases generated is conservative because we do not model the cost of the patients who also have isoniazid mono-resistance and who are often cured with 4HR in the continuation phase, but have an increased rate of treatment failure with 6HE 
. Furthermore, we do not consider additional 4HR-related benefits; TB treatment completion rates in the regimen that is 2 months shorter will likely be higher. Also, if the Ugandan TB treatment guidelines for DOT are fully implemented, the results from our sensitivity analysis suggest that the shorter duration of therapy with 4HR would be expected to result in even greater cost-savings relative to 6HE.
Our results are largely driven by the lower mortality rate as well as the lower failure or relapse rate associated with 4HR. In our model, MDR-TB can only occur in previously treated patients in our model. In sensitivity analyses, we found that even if the MDR-TB rate among previously treated patients were to increase with 4HR (although data from one systematic review suggests higher TB drug resistance rates with 6HE 
), total treatment costs would still be lower with 4HR if the MDR-TB rate were to double because significantly fewer patients would require re-treatment due to the lower relapse rates with this shorter regimen. Furthermore, if the rate of MDR-TB were to increase as high as 30%, 4HR would still be associated with a lower expected mortality rate at an acceptable cost per life saved of $80–$674 depending on the proportion of patients with access to treatment; the mortality rate associated with 4HR for new TB cases is reduced by 2% in HIV-negative patients and 6.1% in HIV-positive patients. Since this reduction applies to all patients that enter the model, it has a disproportionately larger impact on expected mortality than the rate of MDR-TB, which only applies to the subset of patients that require re-treatment.
One of the limitations of our study is that our cost estimates are from the perspective of the Ugandan national health care system with the cost of clinic procedures and cultures based on local rates and a rate of HIV co-infection of 54%. Similar analyses in other countries still using the 6HE regimen should be conducted to understand the generalizability of our results. In addition, our analysis was based on clinical efficacy results as obtained in several randomized controlled clinical trials and cohort studies and actual effectiveness in a real world setting with arguably lower compliance rates could be different. Other limitations include: 1) that the majority of the aforementioned studies were not conducted in Uganda and that these studies did not evaluate whether the numerical differences in mortality between 4HR and 6HE were statistically significant, 2) we did not evaluate the impact of increasing rates of MDR-TB in newly treated patients. Lastly, Green Light Committee (GLC) MDR treatment regimens may allow for significantly lower costs associated with MDR treatment. Although MDR-TB treatment is approved by the GLC for Uganda, drugs are still not available and no patients are currently on treatment through the national program. Finally, despite increased emphasis on cost effectiveness to inform health policy decision-making and coverage decisions, cost effectiveness should supplement, not replace clinical effectiveness evaluations.
In summary, a transition to the strongly recommended continuation phase 4HR regimen is associated with lower costs, lower mortality, a lower overall risk for relapse and, therefore, a reduced need for retreatment. The Ugandan national program will need to continue to enhance health systems to allow for better monitoring of drug compliance to implement rifampicin throughout the course of TB treatment, but the benefit of fewer cases of retreatment and drug resistant TB as well as cost savings, provide compelling arguments for change.