INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In 1978, we first surveyed the tuberculosis chemoprophylaxis practices of the 28 major metropolitan health departments in the United States that reported more than 250 cases of tuberculosis in that year (1). This survey was expanded in 1980 to include treatment as well as chemoprophylaxis practice and, since 1984, the survey has included questions on how health departments formulate tuberculosis treatment and prevention policies (2). The same expanded questionnaire was used in 1988 (4), 1992 (5), and 1996 for this survey-the sixth in the series. Prior surveys have documented the evolution of tuberculosis chemoprophylaxis practices as well as changes in treatment regimens (1), the increasing association between human immunodeficiency virus (HIV) infection and tuberculosis disease (6), as well as budgetary and policy issues among major metropolitan programs (4, 5).

The objective of this investigation was to assess the standards of practice for tuberculosis treatment practice and policy with particular emphasis on changes occurring within the past 4 yr. We found some substantial differences in tuberculosis treatment and prevention practices and policies over the past 4 yr, including changes in treatment regimen, stabilization of the impact of HIV infection on tuberculosis disease, and an encouraging increase in the level of funding allocation to tuberculosis treatment programs from municipal health departments. Our data indicate that many of the challenges that were a major concern in our preceding survey 4 yr ago (5) have been addressed in most metropolitan health departments. This has allowed for implementation of new supervised treatment programs and aggressive programs for prevention that likely account for the decreasing incidence of tuberculosis that has occurred nationally since the prior survey (12).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A written questionnaire was mailed to the designated tuberculosis control officers in the same 28 metropolitan health departments that participated in the 1980 survey, with three exceptions. Between 1980 and 1992, two of the original health departments merged with state departments and were excluded after the merger from the 1992 survey. In this survey, one health department was in apparent transition and could not designate a single person responsible for determination of policy. This department was excluded in the 1996 survey, which was begun in June 1996 and completed in early January 1997. The remaining 25 health departments in this survey all reported more than 250 cases during their initial selection in the chemoprophylaxis survey of 1978 (1). The survey instrument required approximately 35 min to complete and included all questions from the prior surveys of 1984, 1988, and 1992. Additional questions were added in 1992 and in the current survey to determine the impact of: (1) drug resistance, (2) HIV infection, and (3) alternative chemoprophylaxis and treatment programs. These questions focused particularly upon the method by which policy decisions were made and upon budgetary allocations for tuberculosis treatment and prevention. All respondents were assured of anonymity and that the practices of a single department would not be cited explicitly.

Whenever possible, questions were formatted as multiple choice, although respondents were asked to give percentage estimates in some categories. For some questions, an "other" category was included, and respondents were asked for a brief written explanation. We also requested any additional written comments in any category or question. These were readily volunteered, and some are noted in the RESULTS and DISCUSSION sections. All surveys were obtained by written response and, once again, all eligible health departments responded. Programs not responding initially were sent a second survey form, which was followed by phone call within 1 mo if a written survey form was not completed.

We again selected municipal health departments rather than state agencies because of the likelihood of a greater uniformity of policy within a given local agency. Hence, one individual could speak for the policy for care for all patients under the supervision of the agency. Officials were asked to estimate data rather than to survey records. Participants thus were asked to base answers upon overall estimates rather than selective audits of records. This was identical to procedures used in the five prior surveys. In some cases, more than one answer was possible for a query; in others, where answers could not be reliably estimated, the denominator for percentage distribution was adjusted accordingly. To determine prevalence of some treatment regimens across health departments, a weighted average was obtained for each category by multiplying the number of programs employing a regimen by the mean percentage of patients receiving the treatment. This product was divided into a weighted percentage distribution, so that the relative prevalence of a treatment program could be normalized for all programs as percentage of all treatments employed (see RESULTS). Because of the large amount of data collected in this and prior years, this report emphasizes especially areas where important differences have been identified in tuberculosis practice and policy compared with prior years.

Data are reported as the collated survey responses. Where appropriate, data are expressed as mean ± SE. As in prior surveys, no specific hypothesis was tested, and data were based upon perceived policy or estimates rather than calculated statistics by the respondents. Accordingly, statistical analysis was not appropriate in some categories despite some interesting changes in practice between survey years. Such findings are reported in tabular form without statistical testing, and conclusions in the DISCUSSION section are presented accordingly. Where statistical comparisons were appropriate, single comparisons were made by paired t test if the data were normally distributed. Where distributions of data were nonparmetric, chi-square or Wilcoxson's sign rank test was used.

The health departments participating in this survey were Atlanta (Fulton County), Baltimore, Boston, Chicago, Cleveland, Dallas, Detroit (Wayne County), Honolulu, Houston, Jacksonville, Los Angeles, Memphis, Miami, Milwaukee, Nashville, New Orleans, New York, Philadelphia, Phoenix, San Antonio, San Diego, San Francisco, Seattle, Tampa, and Washington, DC.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

All 25 programs completed the survey, and all were included in the survey. Where respondents could not provide the requested information, this response was deleted from the denominator in percentage calculations. Where multiple responses were possible for a single category, the total number of responses in each category was normalized as described above (see METHODS).

Chemoprophylaxis Policies

As in all previous years since 1978, isoniazid remains the primary agent used for chemoprophylaxis in patients with tuberculosis infection, and there is increasing uniformity among health departments in the prophylactic treatment of tuberculosis infection. In 1992, four health departments used an age guideline other than 35 yr (18 to 50 yr). In 1996, all but one health department used 35 yr of age as the upper limit for the administration of chemoprophylaxis in patients not having specific risk factors for tuberculosis disease. The remaining health department used 21 yr of age as the upper limit. In 1992, 81% of health departments used rifampin as an alternative for patients known to be infected by an isoniazid-resistant strain of Mycobacterium tuberculosis. In 1996, 23 of 25 departments (92%) used rifampin (600 mg/d for 6 to 12 mo) as an alternative for chemoprophylaxis; two departments used no alternative drug to isoniazid, and none used any other form of chemoprophylaxis. Application of alternative therapy for chemoprophylaxis, however, varied widely among health departments. Twenty (80%) departments used rifampin in < 5% of all patients receiving chemoprophylaxis for tuberculosis infection. Two departments (8%) used rifampin for as many as 20% of all patients.

Surveillance Policies

Preventive therapy. It is the policy of all metropolitan health departments to evaluate all contacts to index cases of tuberculosis. In contrast to previous surveys, no department indicated a lack of resources to implement this policy in 1996. There was uniform agreement among health departments that all HIV-positive patients would receive tuberculin skin tests (92% of departments), and 19 departments (78%) recommended skin test survey of all patients entering jails or prisons. Although there was a perceived lack of regard for recommendations of the American Academy of Pediatrics (13), 15 of 25 departments (60%) concurred with tuberculin skin testing protocols for children recommended by this organization. Only seven health departments (28%) concurred with the statement that routine skin testing for children should be abandoned in favor of demographically selective testing, and only five departments (20%) believed that only children in high-risk neighborhoods should be tested routinely; this number was virtually identical to 1984, when 19% of departments recommended routine screening of children in high-risk areas only.

Drug toxicity. Despite an increasing number of drugs in tuberculosis treatment regimens (see below), drug toxicity does not seem to be a major concern. There were no specified policies for specific drug monitoring in as many as seven health departments (28%). For patients receiving multiple drug regimens containing isoniazid, 10 departments (48%) monitored isoniazid toxicity clinically and never used laboratory testing for hepatotoxicity. For rifampin, 56% of departments did not perform routine laboratory testing for hepatotoxicity. For ethambutol, all queries related to monitoring for optic neuritis. One department never tested for ophthalmic toxicity for ethambutol unless patients reported symptoms. The remainder tested patients in clinic initially by staff and then either periodically (47%) in clinic or only if symptoms developed (35%). No department used routine evaluation either initially or subsequently (unless symptoms developed or testing abnormalities were apparent in clinic) by an ophthalmologist.

Drug Treatment Policies

Duration of treatment and mode of administration. Between 1980 and 1984, mean duration of treatment decreased substantially from 20.2 ± 2.1 mo to 12.0 ± 1.0 mo, and this decreased further in 1988 to 7.59 ± 1.0 mo (Figure 1). However, in 1992, the mean duration of treatment for tuberculosis disease increased to 9.34 ± 2.32 mo (p < 0.01 versus 1988) following concerns of: (1) decreased budgetary allocations for follow-up subsequent to treatment, (2) increased incidence of HIV-associated tuberculosis, and (3) concerns about drug-resistant tuberculosis. In 1996, there was once more a decrease in mean duration of treatment to 8.00 ± 2.29 mo (p < 0.05 versus 1992) (Figure 1). This appears to reflect both a change in drug regimen now used in 1996, which has resulted in part from increased program allocations for tuberculosis treatment (see below). The distribution of treatment durations among the same health departments in 1992 and 1996 is shown in Figure 2. The predominant difference between 1992 and 1996 is the marked increase in supervised intermittent short-course (6 to 9 mo) regimens. In 1992, only 4.3% of adults patients with tuberculosis received supervised intermittent therapy; in 1996, this number has increased to 45.9% of patients. All health departments now use supervised intermittent therapy in the treatment of tuberculosis for some or all patients.

View larger version (37K): [in this window] [in a new window]   Figure 1.   Estimated duration of treatment. Bars show data for initial treatment for all patients prior to determination of drug susceptibility. In 1992, there was an increase in duration of treatment after two prior substantial decreases. In 1996, there is a decrease in duration of treatment once more, comparable to that estimated for 1988.

View larger version (18K): [in this window] [in a new window]   Figure 2.   Changing treatment regimens between 1992 and 1996. Daily 6- and 9-mo short-course chemotherapy has been replaced substantially by supervised intermittent therapy of the same duration.

Treatment regimens for tuberculosis disease. Although actual estimated practice appears to have differed minimally from departmental policies, some notable changes occurred between 1992 and 1996 in treatment practices, especially among children (Table 1 and 2). In 1992, > 50% of children received isoniazid-rifampin alone as initial treatment for tuberculosis, and only 48.1% received pyrazinamide-containing regimens (Table 1). In 1996, isoniazid-rifampin alone is rarely used in initial treatment for children (3.1%) and pyrazinamide-containing regimens are used in 74.2% of children. Concern about optic neuritis in children also has decreased substantially. In 1992, only 1.3% of children were treated with ethambutol-containing regimens; in 1996, 21.1% of children are treated with ethambutol as part of initial therapy for tuberculosis (Table 1). The predominant change in the past 4 yr in treatment programs for adults has been a continued trend toward the use of multiple drug treatment regimens containing pyrazinamide (Table 2). This corresponds to the increasing use of short-course supervised therapy (see above and Figure 2). In 1992, 18% of all adult patients were treated initially with isoniazid-rifampin alone; in 1996, this has decreased to 6% of all patients. At the same time, there has been a corresponding increase in the number of patients receiving initial four-drug treatment with isoniazid, rifampin, ethambutol, and pyrazinamide (2 mo) followed by 4 to 7 mo of additional treatment without pyrazinamide. In 1992, 81.4% of adult patients received this regimen; in 1996, this is the initial treatment regimen for 94% of patients.

Hospitalization and follow-up policies. All health departments follow some patients after completion of adequate therapy. In 1980, 56% of departments discharged patients with no additional follow-up after completion of treatment. This number decreased to 29% in 1984, and the fraction has remained relatively constant in all subsequent surveys (32% in 1996) (Table 3). The reason for increased follow-up after completion of chemotherapy for tuberculosis disease appears to be related to concerns regarding the decrease in duration of therapy and the increased prevalence of HIV (i.e., concern for recurrence in immunosuppressed patients). However, follow-up has become more selective. In 1992, 44% of departments followed all patients for some period of time after discharge. Routine follow-up was employed only by 12% of all health departments in 1996, and greater emphasis was placed on follow-up of "high-risk" patients after completion of adequate drug treatment. In 1996, five health departments appeared to have no specifically formulated policy for post-treatment follow-up versus prior years when nearly all departments had definitive policies.

As in previous years, inpatient treatment of tuberculosis was reserved for the sickest patients only, and only one program routinely admitted all patients to the hospital for an initial course of treatment (Table 3). This survey did not specify which specific criteria were used for hospitalization. Thus, despite increased incidence of HIV (see below) and continuing primary drug resistance to tuberculosis, ambulatory treatment of tuberculosis remains the predominant mode of therapy in metropolitan health departments in the United States.

Association of HIV and Tuberculosis

In 1980, when the expanded survey instrument first was developed, there was no known association between HIV and tuberculosis. Indeed, AIDS barely was recognized, and the potential association between HIV and tuberculosis had not yet been posited. As late as 1988, tuberculosis was cited only by seven health departments as a cause of increased incidence of tuberculosis. (These were seven of the nine health departments reporting an increase of tuberculosis during the preceding 4 yr.) In 1984, only 2.54% of patients having tuberculosis were thought to have HIV as a related factor. This increased to 17.8% in 1992, with eight health departments estimating that between 20 and 40% of all new cases of tuberculosis were infected with HIV. In 1996, this profile remains virtually unchanged, and the increased incidence of HIV-associated tuberculosis between 1984 and 1992 now may have plateaued (Figure 3). In 1996, in metropolitan health departments in the United States, 18.0% of all patients were estimated to have HIV-associated tuberculosis, and the frequency distribution remains approximately the same compared with 1992 (Figure 3).

View larger version (16K): [in this window] [in a new window]   View larger version (27K): [in this window] [in a new window]   Figure 3.   Estimated incidence of HIV-associated tuberculosis in major metropolitan health clinics in the United States. (Upper panel ) Mean annual incidence for all health departments from 1984 to 1996. After a sharp increase in HIV-associated tuberculosis between 1984 and 1992, there appears to be a leveling effect in the past 4 yr. (Lower panel ) Estimated frequency distribution of HIV infection associated with tuberculosis in 1984 to 1996. The "rightward" shift also appears to have been halted between 1992 and 1996.

Budgetary Allocations

After years of diminishing or stagnant funding for tuberculosis treatment and prevention, there has been a dramatic increase in dollars after correction for inflation allocated to tuberculosis treatment versus the prior 4 yr (Table 4; Figure 4). In 1988, there was an approximate 8% decrease in funds allocated for tuberculosis treatment versus 1984. This allocation did not change after correction for inflation during the next 4 yr. However, in 1996, there was a dramatic increase of 83.9 ± 32.7% in the mean allocation of funds for the treatment and prevention of tuberculosis (Figure 4). Although there was substantial variation among departments, only two departments reported a decrease in inflation-corrected allocations (11.0 ± 8.5%), while 15 departments reported substantial increases. In the cases where budgetary allocations were decreased, this was attributed to local mandates (i.e., city or county budgetary actions). In prior years, both state and federal, as well as local, allocations were attributed to decreases in tuberculosis treatment resources. In subjective comments solicited through the survey, many tuberculosis control officers now indicated that resources were sufficient to meet the desired objectives and that some policies had been altered accordingly. The increased use of supervised intermittent drug therapy, which is more labor-intensive than self-administered treatment (but also more reliable), was directly attributed in many cases to increased resources for treatment.

View larger version (20K): [in this window] [in a new window]   Figure 4.   Mean change in budgetary allocation for treatment of tuberculosis among metropolitan health departments in the United States. There was no significant change in allocation in 1988 or 1992, despite increasing caseloads and an increasing national incidence of tuberculosis (see DISCUSSION). An encouraging trend is seen in 1996, which, however, still is not uniform among all health departments (see Table 4).

Policy Decisions

As in prior years, the policies of the Centers for Disease Control (CDC) continued to have the greatest impact on the formulation of metropolitan health department policy for the treatment of tuberculosis. On a scale of 1 (greatest impact) to 4 (least impact), CDC recommendations were scored 1.13 ± 0.07 (Table 5). This was nearly identical to the impact scores obtained in 1988, when this first was queried, and also in 1992. State health departments were next most influential in the formulation of policy in metropolitan health departments (2.25 ± 0.18). By contrast, policies of the World Health Organization (WHO) were weakly regarded (3.62 ± 0.59). As in prior years, the recommendations of the WHO were regarded as least influential in a formulation of tuberculosis treatment policy.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This is the sixth in a series of surveys to determine the treatment practices and policies of metropolitan health departments in the United States. Since 1980, we have used a stable base of questions that has allowed us to assess changes and trends in treatment of tuberculosis that may occur even within a relatively short 4-yr period. For example, the mean treatment time for tuberculosis between 1980 (> 20 mo) decreased to 12 mo by 1984 (Figure 1); we were able to determine that this was in large part due to adoption of rifampin-containing regimens that allowed for shorter times to achieve cure for tuberculosis disease. Other patterns have emerged more slowly, e.g., the recognition of the need by agencies funding metropolitan tuberculosis programs that increased allocations for treatment and prevention were necessary. Some major trends in tuberculosis treatment and prevention between 1980 and 1996 are summarized in Table 6.

Although there remains a relatively high degree of conformity of practice by metropolitan health departments in some areas (e.g., chemoprophylaxis and preferred drug regimens) (14), this consensus has evolved substantially over the 18 yr that we have conducted this survey. Isoniazid still is the overwhelming choice for prophylactic treatment of tuberculosis infection. Throughout the years, the 35-yr-of-age guideline for "routine" chemoprophylaxis with isoniazid has remained the standard in nearly all cases. However, in recent years, the introduction of rifampin as an alternative treatment has been adopted widely and, in 1996, is used by all but two of 25 metropolitan health departments. Although used infrequently by metropolitan health departments, this regimen has been shown to be effective where required (15), and adoption of this policy, which initially was based on consensus prior to clinical study, is likely to have a significant impact in preventing tuberculosis infection from progressing to disease.

In contrast to chemoprophylaxis practices, drug toxicity is not uniformly monitored by metropolitan health departments. For some drugs, as many as 25% of health departments have no recommended monitoring policy. For drugs causing hepatotocixity, e.g., isoniazid and/or rifampin, many health departments do not follow guidelines previously suggested for routine laboratory monitoring but rather follow clinical parameters and obtain biochemical testing only in symptomatic patients. In previous years, limited resources mandated limited monitoring for drug toxicity (4, 5, 16, 17). However, this survey indicates that, in 1996, allocations for tuberculosis treatment have increased substantially (see discussion below). Nonetheless, biochemical monitoring for drug toxicity has not changed, indicating that, even with increased financial resources, routine biochemical monitoring is not perceived as a high priority for tuberculosis treatment programs. These data suggest a possible need to establish clearer and more uniform recommendations for drug toxicity monitoring that are congruent with established therapeutic/toxicity relationships. Similar decision analysis has been developed for predicting likelihood of active tuberculosis (18).

Treatment regimens in 1996 have changed substantially in: (1) allocation of multi-drug regimens and (2) utilization of short-course supervised chemotherapy (19). This survey marks the disappearance of "conventional" 18-mo chemotherapy of tuberculosis disease with isoniazid-rifampin-streptomycin or isoniazid-ethambutol. All programs now use shorter-course therapies of 6 to 9 mo in duration, except under "special" circumstances that usually are related to poor compliance or initial drug resistance. In 1996, it is estimated that > 90% of all patients are treated initially with pyrazinamide-containing regimens, and nearly half of all adult patients receive supervised chemotherapy (Figure 2). Treatment for children also has changed substantially. Ethambutol, which previously was not used because of concern for ophthalmic toxicity, now is used approximately 20 times as often in children, and the use of pyrazinamide-containing regimens for children has increased by approximately 50% (Table 2). Possibly because of the remarkable increase in supervised therapy and use of multiple drug regimens, the duration of treatment, which increased from approximately 7.5 mo to > 9 mo between 1988 and 1992, now has decreased to 8 mo (Figure 1).

There is additional cause for encouragement. The estimated incidence of HIV-associated tuberculosis, which has increased steadily since 1984, now appears to be stable at approximately 18% during the past 4 yr. In 1992, we raised a substantial concern in this survey that tuberculosis programs were substantially underfunded to meet the health care needs of metropolitan clinics (5). Case loads were increasing, as was the incidence of primary drug resistance and HIV-associated disease (22, 23), while revenues for treatment programs were decreasing. These concerns were echoed simultaneously in the lay press (24, 25). Between 1992 and 1996, there has been an impressive increase in the budgetary allocations for treatment of tuberculosis. While some departments still continue to experience decreased budgetary allocations in inflation-corrected dollars, real growth on average has increased almost 85% (Figure 4, Table 4) for the 25 metropolitan health clinics, and many tuberculosis control officers no longer perceive a problem in budgetary resources. Because this substantial increase in funding for tuberculosis programs was not anticipated, we were not able to determine the source of increased funds. As we know of no new categorical federal funding programs that are uniformly applicable to metropolitan health departments, it is likely that increased funds have been obtained from state or local budgets. These increased funds are being allocated to supervised treatment programs, which in turn has reduced the duration of treatment and the need for routine post-treatment follow-up of patients deemed not to be at high risk (Table 3). In addition, a broader community outreach for screening programs now is possible and appears to be in the process of implementation. However, six departments still have similar or decreased funds for treatment and prevention programs compared with 1992. Thus, the apparent improvement in funding for tuberculosis control is not universal. Another concern is the maintenance of these increased funding levels. Where increased funds have become available, this has resulted in expanded use of directly administered therapy and increased tuberculosis surveillance practices, which likely are related to the decrease in tuberculosis cases since the prior survey. Should funding levels decrease in the future, it is likely that progress made recently toward tuberculosis eradication will not continue.

It is important to note the major limitation of our findings. While many questions related to policy rather than practice, the large amount of data requested of each respondent required estimation rather than calculation for determination of statistical survey questions. As we have noted in prior publications (2), these estimated data have closely paralleled national trends as calculated by the Division of Tuberculosis Elimination, Centers for Disease Control and Elimination. In recognizing that our epidemiologic data are not precisely quantitative, we also note that the continued resurvey of the same health departments likely reflects the changing perception of tuberculosis control officers who face the challenge of management of tuberculosis in urban settings. When averaged over 25 major health departments during a period of > 18 yr, it is likely that these data closely reflect practices within metropolitan health clinics as they are evolving in the United States (12).

We conclude that tuberculosis treatment in the major metropolitan health clinics in the United States consists predominantly of short-course therapy and that there is a growing trend to supervised therapy containing an initial course of pyrazinamide. We find that nearly all programs have adopted alternative chemoprophylaxis programs for isoniazid-resistant infections that use rifampin, although utilization of these programs is not a substantial need. We are especially encouraged to report the first increase in funding for tuberculosis treatment since 1984, when we first asked this question. The improved allocation of resources has been substantial in many programs, and we believe that this is likely to be a substantial factor in the reduced incidence of tuberculosis in the United States.