Critically Appraised Topic

Question: Should we treat patients older than 35 who had a significant positive PPD, but at low risk for TB activation, with INH?

Appraiser: Eric W. Terman, MD

Date: 1/22/98

Clinical Bottom Lines:

1. It appears that treating patients, older than 35 and at low risk for TB activation, with INH is both life saving and cost effective, but they need to be monitored to prevent death by INH induced hepatitis.

2. This study only broadens who we treat with INH. The current recommendations are, everyone younger than 35, and anyone older than 35, who are at high risk for activation, e.g. diabetics, alcoholics, immunocompromised patients, patients with abnormal chest x-rays.

3. Situations where starting INH needs to be re-thought include: patients who are non-compliant and DOT is not available, pre-existing liver disease where monitoring for signs and symptoms of INH toxicity is extremely difficult.

Methods:

This was a cost effectiveness and decision analysis article. They initially constructed a decision tree that included the possible outcomes for each intervention and the multiple possible outcomes. For example they looked at how often people would die from INH, versus those who would be treated and still get TB versus those who would be treated and do well and not get TB. After their model was constructed they found probabilities from the existing literature on rates of activation, death, cost of hospitalization, etc. From these they plugged it into a computer, which gave an answer. Using the computer they simulated a cohort of patients to see what the results were. The results included cost savings and years of life saved, but they were averaged over the entire cohort. They finished off their study by changing the baseline probabilities to see if changing the baseline data altered the conclusions. (This is called a sensitivity analysis).

Results:

Results of the Markov Model

	35-49 years old		50-69 years old		>70 years old
	no isoniazid	isoniazid	no isoniazid	isoniazid	no isoniazid	isoniazid
NNT to prevent one death		529		565		654
Prob of survival to 1 yr.	0.99809	0.99811 p=0.00002	0.99529	0.99533 p=0.00004	0.97371	0.97383 p=0.00012
Prob of death from TB	0.00269	0.00079	0.00252	0.00073	0.00218	0.00063
Costs per patient	290.81	190.24	246.50	177.25	163.66	153.04

Results of the Sensitivity analysis, the values when the conclusions are no longer upheld

	baseline occurrences	cross-over when the current conclusions would be not be supported	difference required to alter conclusions
prob of fatal hepatitis	0.002%	0.083%	35 times greater than current prevalence
efficacy of INH	0.85 for 6 mos (0.71-0.98) 0.15 for 3 mos (0.1-0.3)	0.1	a factor of 7x less efficacious, using the lower efficacy number for 6 mos.
Prob of TB activation	0-9 yrs 0.11% 10-14 yrs 0.08% >15 yrs 0.06%	0.018	factor of 3 to 10 depending on how long the patient has been PPD positive
cost of TB per case	In this analysis it was always more cost effective to treat TB secondary to the costs of subsequent hospitalization and treatment costs associated with active TB

Comments:

1. This was a well done study the stated its assumptions and showed the data that were used to generate the results. The decision tree was constructed well.

2. The results included in the table above did not include the results when they examined the cost of cases caused by an index case that was not treated. These results were even more convincing.

3. As TB becomes more prevalent, in order to prevent the disease from becoming endemic, it will be even more important to treat even those people who are at low risk for having active TB.

4. The original recommendations were the result of assuming a higher rate of hepatic failure than actually occurs. This would indicate that all PPD positive patients should be treated.

Source:

Salpeter SR, Sanders GD, Salpeter EE and Owens DK. Monitored Isoniazid Prophylaxis for Low Tuberculin Reactors Older Than 35 Years of Age: A Risk-Benefit and Cost-Effectiveness Analysis, Annals of Internal Medicine. 127(12):1051-1059.