INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Preventing tuberculosis (TB) has been one of the cornerstones for eliminating tuberculosis for over 40 yr. The efficacy of treating latent tuberculosis infection (LTBI) first with daily isoniazid (INH) and para-aminosalicylic acid for 18 mo (1) and then with 12 mo of daily INH in the U.S. Public Health Service trial (2) was established and subsequently confirmed by other studies (3). INH was suitable because it was safe, cheap, easy to take, well tolerated, and effective (1, 3).

There were problems with INH treatment of LTBI, however. Hepatitis was reported mostly in persons older than 20 yr (9). Another problem was compliance. Poor compliance was noted in one early study (2) and some later studies reported similar observations (6, 15). There was a relationship between compliance and outcome (8). Wobeser and coworkers reported appallingly poor compliance in Canadian Plains Indians who took self-administered treatment for LTBI (SAP) with daily INH for 1 yr (15).

To improve the compliance of treatment for LTBI, a directly observed prophylaxis (DOP) with twice weekly INH and rifampin (RMP) was considered because of the reported efficacy of 6 mo daily INH and RMP prophylaxis in children (16), the efficacy of RMP as a sterilizing agent (17), particularly when given with INH (17, 18), and the suitability of RMP for intermittent therapy (19).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

From January 1, 1992 to December 31, 1995 infected aboriginal patients, who were eligible for treatment of LTBI and agreed to start, began DOP with INH and RMP twice weekly for 6 mo. They were followed to December 31, 1997. The dosage for INH was 15 mg/kg to a maximum of 900 mg/d and for RMP was 10 mg/kg to a maximum of 600 mg/d. Prescriptions were written monthly and sent in unit doses to the TB worker living in the community. The TB worker delivered one dose twice weekly to the patient and observed and recorded each dose. A medication audit was obtained from the records of the TB workers, from a count of empty unit doses that were returned monthly to the TB Pharmacy, and from monthly mobile clinic reviews by TB consultants. TB workers were trained by the TB Program nurses, funded by the TB Program, hired by the community, and supervised by the Community Health Nurse. There was close communication with the TB Program nurses through telephone, fax, and mobile clinics.

Compliance was calculated from the ratio doses taken/doses prescribed. Treatment continued for 6 mo or until a minimum of 90% of 52 doses had been taken (range 6-9 mo, mean 6.4 mo). Patients who discontinued treatment of LTBI before completion because they did not wish to continue were classified as decision to quit. If they could not be found for more than 1 mo during treatment they were classified as lost to follow-up. Liver function tests were performed only when there was clinical suspicion or high risk of hepatitis (20).

For comparison with the DOP group, the records of all aboriginal patients who began self-administered daily INH for 1 yr from January 1, 1986 to December 31, 1989 were reviewed. They were followed to December 31, 1991 to standardize the observation period. The dosage for INH was 15 mg/kg to a maximum of 300 mg/d.

Incipient tuberculosis, cases that developed while on treatment for LTBI, was also recorded. The group data including mean age in years and follow-up period in patient-years are listed in Table 1. The epidemiological situation for the treatment periods 1986-1989 and 1992- 1995 is listed as the case rate/1,000/yr. The interval from skin test to start of treatment was recorded to ensure that this interval did not exceed the period in which the main risk of TB had passed, namely 2 yr (21, 22). There were no patients with known human immunodeficiency virus (HIV) infection.

In those patients who developed TB, the diagnosis was established bacteriologically or clinically. In the bacteriological cases, all patients had organisms sensitive to antituberculosis drugs. The specimens were cultured for Mycobacterium tuberculosis using the Bactec 460 TB system (23) and identified using a DNA probe (23). In the culture-negative cases, the diagnosis was based on assessment including skin test, chest radiograph, and response to treatment (24).

Drug toxicity was monitored by the TB worker twice weekly, and by the nurses and TB consultants monthly. The reasons why patients stopped drugsdeath, pregnancy, toxicity, decision to quit, or were lost to follow-upwere recorded. To account for the significant proportion of patients who did not complete SAP, the toxicity rate was calculated by using patient-years of taking drugs as the denominator.

The patients in the two treatment groups who discontinued treatment for any reason were recorded at monthly intervals using a Kaplan-Meier plot and compared at 6 mo using the Mantel-Haenszel test (25). The compliance for the two treatment groups was compared using the chi square test. The completion rate and side effects for the two groups were compared using the chi square test. The rate of tuberculosis following treatment was compared using the Fisher exact test. The age of patients and the duration of follow-up in the two groups were compared using the t-test. The aboriginal rate of tuberculosis from 1986 to 1991 was compared with the rate in 1992-1997 using the chi square test.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Of patient records, 403 for the SAP group and 591 for the DOP group were reviewed. The SAP patients were older, 14.5 ± 11.7 yr, compared with the DOP patients who were 8.7 ± 9.0 yr old (t test, p < 0.001). The SAP patients were followed for 4.0 ± 1.2 patient-years, compared with 3.6 ± 1 patient-years for DOP patients. The DOP patients were followed for a shorter mean period (t test, p < 0.001) while having a higher number of total patient-years. The number of patients by age is listed in Table 2. Nine patients developed tuberculosis after completion of treatment and six patients developed (incipient) tuberculosis during treatment by December 31, 1991 in the SAP group and one patient developed tuberculosis after treatment and one developed (incipient) tuberculosis during treatment in the DOP group by December 31, 1997. When incipient tuberculosis was included, the rate of tuberculosis was 9/1,000 patient-years in SAP patients compared with 0.9/1,000 patient-years in DOP patients. The difference was significant, p < 0.001 (Fisher exact test). During the study period the overall rate of TB in the aboriginals did not changethe 1986-1989 rate was 1.37 new cases/1,000/yr compared with the 1992-1995 rate of 1.21 new cases/1,000/yr, p = 0.32 (chi square) (Table 1).

Most SAP patients stopped treatment prior to 6 mo when 46% remained on follow-up; 19% remained on follow-up for the prescribed 12 mo (Figure 1). Of DOP patients 82% patients were on follow-up for the prescribed 6 mo. There were significantly more DOP patients on follow-up at 6 mo compared with SAP patients, p < 0.001 (Mantel-Haenszel test). The resulting compliance for DOP patients was higher, 84 ± 21%, than for the SAP patients, 24 ± 17%, p < 0.001 (chi square) (Table 1).

View larger version (12K): [in this window] [in a new window]   Figure 1.   Kaplan-Meier plot showing the probability of patients continuing treatment by month of observation. (Squares) 6 mo twice weekly DOP; (triangles) 1 yr daily SAP. The difference at 6 mo was significantly higher for DOP, p < 0.001 (Mantel-Haenszel test).

Table 3 lists the reasons for not completing treatment of LTBI. The decision to quit and loss to follow-up were the main reasons for stopping treatment in both groups but was significantly lower in the DOP group than in the SAP group, p < 0.001 (chi square). Side effects as a reason for not completing treatment was higher in DOP patients than in SAP patients, p < 0.001 (chi square). Table 4 shows that the rate of side effects/1,000 patient-years of taking drugs rose with age, p < 0.001 (chi square) and was higher in DOP patients for each age group, p < 0.05 (chi square). The DOP patients reported higher rates of gastrointestinal symptoms (mostly nausea), neurological symptoms (mostly headache), and miscellaneous symptoms (mostly fatigue), p < 0.001 (chi square) compared with SAP patients. Skin allergic reactions, p = 0.51 (chi square), and hepatitis, p = 0.82 (chi square), were similar in both groups (Figure 2). Side effects did not lead to hospital admission or death in either group. Paresthesia was recorded in one SAP patient. The mean interval from the first positive skin test to the start of SAP was 5.1 ± 2.3 mo and to the start of DOP it was 5.5 ± 1 yr (Table 1). Therefore both groups started within the 2-yr period of greatest risk.

View larger version (22K): [in this window] [in a new window]   Figure 2.   The rate of side effects/1,000 patient-years of taking medication. Patients in the DOP group had significantly more side effects compared with patients in the SAP group, p < 0.001 (chi square). The differences were gastrointestinal (GI) (mostly nausea), miscellaneous (misc) (mostly fatigue), and neurological (neuro) (mostly headache). The rate with skin allergic reactions, p = 0.51 (chi square) and hepatitis (hep), p = 0.82 (chi square) was similar in both groups. SAP, dark bars; DOP, light bars.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Intermittent DOP was implemented to improve the outcome of the treatment for LTBI. It did improve the outcome with some increase in non-life-threatening side effects. The two-drug twice-weekly regimen was desirable because RMP was an effective sterilizing agent particularly when given with INH (17, 18), a suitable drug for intermittent therapy (19), and the regimen was affordable with 52 doses. There was an added benefit because of a 4% prevalence of drug resistance to INH in Saskatchewan aborigines (26), in whom INH alone risked failure of the treatment for LTBI.

The improved outcome of DOP was mainly due to improved compliance (Table 1) and completion of treatment (Figure 1). The effectiveness of daily INH for 6 mo and 1 yr has been established and relates to compliance; 1 yr of daily INH showed 70-93% protection with 80+% compliance but only 49-75% protection with 60-68% compliance (6, 8). This compares with 69% for 6 mo of daily INH alone with 80+% compliance (8). However, the effectiveness of this regimen in the 1986-1989 group of patients was unsatisfactory mainly because of failure to take most of the prescribed doses. Only 46% of SAP patients were taking INH at 6 mo compared with 82% of DOP patients who were taking INH and RMP at 6 mo. DOP was successful in keeping patients on treatment.

There were several factors that affected the rate of tuberculosis in infected patients. These included ethnic origin, age, follow-up period, and the interval from the time of infection to intervention, because about 50% of the lifetime risk occurred within 2 yr following infection (21, 22). The two groups were similar in risk for tuberculosis as the ethnic susceptibility was standardized by including only aborigines, age was standardized by comparing age groups, and the follow-up period was standardized with a 6-yr observation period from the start of treatment. Although the mean follow-up period for DOP patients was significantly shorter than for SAP patients, patient-years of follow-up for DOP patients was higher than for SAP patients with a bias against the DOP outcome. The interval from infection to intervention was more problematic as the time of infection was not known in some cases, especially in patients 10 yr of age and older. The older the patient the less likely the first positive skin test coincided with the time of infection. The bias of varying intervals was diminished by age stratification and by showing that the interval was less than 6 mo in both groups.

Another bias was the time interval between the treatment groups, an inherent problem in an uncontrolled before and after study design. These related to potential changes in the epidemiological picture and in the TB Control program. The rates of tuberculosis in the aborigines were similar in the two intervals, suggesting that the epidemiological situation did not change (Table 1). The method of detection of tuberculosis was mainly through symptoms for both groups. However, the proportion that was detected through symptoms diminished from 78% in 1986-1989 to 62% in 1992-1995 (26). It showed a shift to more screening/active case finding. Although this did not increase the number of reported cases, it diminished the likelihood of missing cases (Table 1).

Eighty-two percent of the DOP patients and 19% of the SAP patients completed treatment. The main reason for discontinuing treatment was a decision to quit by 304 of 403 (75%) SAP patients compared with 53 of 591 (9%) DOP patients (Table 3). The difference with DOP patients was probably due to encouragement that was offered by TB workers during regular twice weekly contact and arranging drug delivery to fit the patient's schedule compared with one to three monthly scheduled visits to the clinic. The decision to quit in both groups was higher than previously reported (13, 27). This reflects the realities of drug delivery in this population and locale.

The second most common reason to stop treatment was side effects39 side effects/1,000 patient-years of drugs for the SAP patients and 136 for the DOP patients. DOP patients reported higher rates of non-life-threatening symptoms such as nausea, fatigue, and headache. Although DOP patients showed a higher overall rate of stopping drugs, the rates of stopping drugs due to more serious side effects such as skin allergic reactions and hepatitis were similar (Figure 2). In spite of side effects, the completion rate of preventive treatment in DOP patients was superior to SAP patients.

There were some reasons to suggest that side effects in the SAP patients were underreported. Of SAP patients 2.2% reported side effects compared with 6.4% and 5% in earlier reports (12, 13). This difference remained when stratifying for age (13). Second, the rate was probably underestimated in the SAP patients because of the large number of patients who defaulted. They may have had side effects but did not report them. Third, the TB workers were trained to look for side effects in the twice weekly contact. Therefore, the likelihood of finding symptoms due to side effects was higher for DOP patients.

Several reports have appeared for combined daily RMP and INH treatment of LTBI in a variety of patient groups (16, 28, 29). Protection rates varied from decline in cases over time with 6 mo daily therapy in children (16), to 36% with 3 mo of daily therapy in adults with silicosis (28), to 100% with 4 mo daily therapy in homeless adults infected with INH-resistant organisms (29). Side effects were not reported (16) or were as high as 53% (19% when corrected for side effects reported by controls) (28) leading to cessation from 5% to 16% of patients (28, 29). These reports with daily INH and RMP showed that drug intolerance rose with age, that patients tolerated INH and RMP as well as INH alone, and that there was a high level of protection in patients who took 4 mo or more of therapy. Daily RMP alone for 6 mo resulted in a probable protection of 85% in a group of adolescents who were exposed to INH resistant tuberculosis (30). Twenty-six reported side effects, because of which 4% stopped treatment. An intermittent regimen with 18 mo of three times weekly directly observed INH and ethambutol in Eskimo children and adults reported 100% protection (31). Treatment was stopped because of side effects in 1% of these patients.

A trial of RMP and INH, and RMP, INH, and pyrazinamide (PZA) treatment of LTBI in a 3-mo daily regimen in an HIV- and TB-infected Ugandan population (32) reported 61% protection with RMP and INH and 50% protection with RMP, INH, and PZA. Of patients 9.7% taking RMP and INH reported side effects, of whom 2.3% stopped treatment, whereas 24.7% of patients taking RMP, INH, and PZA reported side effects, of whom 5.6% stopped treatment. Three-month regimens may be effective in HIV⁺ populations.

The main obstacle to drug delivery was patient life-style. This was diminished by harmonizing drug delivery into the patient's routinebedtime dosing for drowsiness, evening dosing if work schedule required it, holding medication up to 2 wk when fire-fighting or trapping, flexible clinic hours, taking time for questions, coordinating solutions for child care, eye glasses, other doctor appointments, and providing transportation where needed. The structure in which the TB worker functioned required some persistence to ensure that the TB Program received regular, timely, and accurate records of DOP. This difficulty was diminished with regular communication by telephone, fax, meetings with supervisors, and mobile clinics.

Was the cost/benefit ratio of DOP acceptable? A previous Canadian review of treatment costs for 1 yr of daily INH was estimated to be $132 to $7,564 (median $875) and for a case of tuberculosis in a 20-yr-old patient from $7,906 to $41,386 (33). The 1995 cost of treatment of LTBI and TB treatment based on the consumer price index that rose 29% (34) during the interval was $170 to $9,758 and $10,199 to $53,388, respectively. The additional cost of DOP was estimated at $750 1995 dollars per patient or $443,250 for 591 patients. Because the rate of tuberculosis was constant over time, 22 cases of tuberculosis were expected in the 591 patients if SAP had been used. Only two cases were reported. The benefit of 20 prevented cases of tuberculosis was $203,980 to $1,067,760. The cost benefit ranged from 0.5 (50 cents benefit/ dollar cost of TB prevented) to 2.1 (210 cents benefit/dollar cost of TB prevented). Because contacts to cases, drug-resistant cases, and deaths were not considered in these estimates, the range of the cost/benefit ratio was lower or less costly. Additional benefits in this remote population included fewer hospital stays far away from home in a different cultural setting. The conclusion was that DOP was desirable and affordable.

A regimen of 52 doses of twice weekly INH and RMP was an affordable, effective, and well-tolerated regimen to improve the outcome of the treatment of LTBI in aboriginal patients with high rates of tuberculosis. The better outcome was mainly the result of improved compliance and completion rates.