CDC Logo Tuberculosis Information CD-ROM   Image of people
jump over main navigation bar to content area
TB Guidelines
Surveillance Reports
Slide Sets
TB-Related MMWRs and Reports
Education/Training Materials
Ordering Information


U.S. Department of Health and Human Services


This is an archived document. The links are no longer being updated.

TB Notes 1, 2002

Highlights From State and Local Programs

Housing Homeless TB Patients in California

Homelessness has long been recognized as an important predictor of poor adherence to TB therapy. During the years from 1994 to 1996, the California Department of Health Services (CDHS), TB Control Branch (TBCB), received Report of Verified Case of TB (RVCT) forms that identified an average of 339 (7.36%) TB patients as having been homeless each year.

Homeless patients are among the most difficult to cure because they are often difficult to locate. In order to ensure that patients could be located for the administration of consistent directly observed therapy (DOT) and thereby complete therapy within 12 months, TBCB decided that a housing component should be added to the TB control program.

In 1996, at the behest of the California Tuberculosis Controllers Association and the California Conference of Local Health Officers, TBCB proposed a change to the Governor's budget that added $2.9 million for fiscal year 1997/98 for a program addressing the needs of homeless and unstably housed TB patients in California. The funds were also to be used to detain persistently nonadherent patients. TBCB allocated housing funds to local health jurisdictions (LHJs) by a formula based on the number of homeless patients they had reported.

Soon after the inception of the program, it became clear that providing housing alone was not the solution. In addition to housing, patients needed food, transportation, and other services. Some homeless patients who did not want housing could be kept on treatment by being provided with other incentives and enablers. Finally, it was determined that other solutions were needed to prevent patients from becoming homeless in the first place, and the problems that lead to homelessness should be addressed. Providing additional funding for other compliance-enhancing measures and for less restrictive alternatives to detention might reduce the need for detention and improve completion of treatment (COT).

Therefore, about 6 months into the program, TBCB broadened the scope of services provided to include compliance-enhancing measures and less restrictive alternatives to detention. The branch also extended the target population served by this program to include persons at risk of becoming homeless as a result of their TB diagnosis. TBCB increased services to homeless patients who refused housing in an attempt to stabilize their lives enough to improve compliance with DOT.

Broadening the scope of services provided, however, created a need for additional personnel at the local level to provide these services. Not only were homeless patients provided with shelter and food; they now received increased DOT, case management, and social services. Since FY 1999, approximately 47% of the total housing budget allocated to the LHJs is currently spent for personnel who provide direct services to clients. It supports 22.7 positions statewide (13.1 outreach, 0.9 field investigation, 6.0 case management, 2.5 social work, and 0.2 miscellaneous; numbers rounded to one decimal). The remaining 53% of the housing budget is spent on food and shelter.

BCB collected data over the 2-year period of January 1, 1998, through December 31, 1999, using a data collection instrument called the Report of Homeless Tuberculosis Patient or Suspect (RHTP). Local health departments were asked to complete an RHTP on each patient with known or suspected TB who was homeless within the 12 months prior to diagnosis, at the time of diagnosis, or at any time prior to the final disposition of treatment, whether or not housing was provided. The purpose was to ascertain the outcome of housing and treatment for every homeless TB patient or suspect identified in the state during that period. The RHTP collected information on whether or not the patient had been offered and accepted housing, how long the patient was actually housed, the type and duration of housing provided, and the reasons for which the patient left housing. By linking the RHTP to the RVCT for the same patient, the demographic characteristics and the outcome of therapy could be correlated to the housing provided. Additionally, the RHTP was linked to the system by which LHJs invoice the State to determine the cost per day and per patient for providing the housing.

TBCB received 620 RHTPs of suspects and cases over this 2-year period for both homeless cases and suspects. Of the 620 patients, 327 (52.7%) were offered housing. However, when the analysis excludes one jurisdiction in which only 24.2% (70/289) were offered housing, the percentage for the remainder of the State jumps to 73.1%.

The branch examined RHTPs to determine the reasons for which the remaining 293 patients were not offered housing; we were able to determine the reason for 228 of these. Of the 228, 144 (63.1%) actually had some form of shelter, although technically homeless because they had no official address. TB was ruled out in another 35 (15.4%) patients. An additional 38 either moved, died, or were lost to follow-up, leaving only 11 (4.8%) patients who were not offered housing for undetermined reasons.

Of those offered housing, 92.3% were actually housed. In fiscal year 1998/99, the only year for which we had complete fiscal data, 263 patients were housed for 19,078 days at a cost to CDHS of $586,538, an average of $2,230 (73 days) per patient and $30.74 per day of housing. We were unable to determine the amount of local funding spent on housing or the number or additional patients that may have been housed using local funds.

The majority (73.8%) were housed in single room occupancy (SRO) hotels. An additional 21.4% stayed in "other" facilities. The category "other" includes apartments, homes, and some inpatient facilities. Only a very few patients were housed in YMCAs (<1.0%), rehabilitation centers (<1.0%), or board and care facilities (4.1%).

Thirty-nine patients were offered housing, but were not actually housed using CDHS funds. Of these, 7.7% refused housing, 46.2% were actually housed by another program, and 18% were detained, incarcerated, or hospitalized. Treatment was terminated for one patient, leaving 10 (25.6%) patients who were not housed for various other or unspecified reasons.

We received follow-up RHTPs on 243 (84.3%) of the 288 patients housed by the program. Of these, 87 (35.8%) patients left housing because their treatment was terminated. Another 93 (38.2%) either moved to or were detained in other facilities. They therefore remained housed for the continuation of their treatment.

We attempted to determine the treatment status of each patient at the time housing was terminated. As mentioned above, over 35% left housing because treatment was terminated. Another 22.2% left housing but continued treatment in the community and 22.6% left because TB was ruled out as a diagnosis. Only 19.3% left because they moved out of the area or died, or for unknown reasons. It is important to remember that these figures do not necessarily reflect the final outcomes of these cases.

How well does the RVCT capture the extent of homelessness among TB cases?
To better understand the magnitude of the problem, we compared RHTP and RVCT data for 1998, the year for which RHTP data were most complete. Together, these two data sources identified 287 homeless TB patients, 45 (16%) of whom were not identified as homeless on their RVCTs. This suggests that the RVCT may underestimate the number of homeless persons with TB by 16%. A possible reason is that incarcerated patients are not usually reported as homeless, but may actually be homeless when released from jail. Nor does the RVCT system capture the number of patients who become homeless subsequent to diagnosis.

We determined patient characteristics of 585 persons with TB reported as homeless on either the RHTP or RVCT in 1998/99. Eighty-two percent of homeless persons were men, and 38% were foreign-born persons. Fifty-eight percent were non-Hispanic white persons, 29% were black persons, 8% were American Indians, 2% were Asian/Pacific Islanders, and 2% were unknown. In terms of substance abuse, 15% of homeless patients reported injecting drug use, 30% noninjecting drug use, and 49% excess alcohol use in the past year. The RHTP captured homeless patient demographics that were significantly different from those captured on the RVCT. This comparison suggested that more American Indian, Hispanic, and foreign-born persons may be homeless prior to or at the time of diagnosis than was previously thought.

Trends in treatment outcomes for homeless persons with TB identified on RVCT. We also looked at treatment outcomes for homeless patients (as identified on the RVCT) before and after the implementation of this program. The analysis is based on the most recent follow-up RVCT available, and is restricted to patients who began treatment and had no resistance to rifampin. The rate of completion in less than 12 months had already risen from 57.4% in 1995 to 66.3% in 1996 and then to 69.2% in 1997. In 1998, the first year of the program, the rate of completion in less than 12 months fell to 66.0%, but rose again to 69.3% in 1999. The rate of completion in over 12 months fell from 13.8% in 1995 to 8.9% in 1996 and 8.4% in 1997, rose to 12.9% in 1998 and then finally fell to 7.5% in 1999. Similarly, the rate for the category of "died" fluctuated between 6.3% and 10% in these years.

Categories for outcomes that may have been influenced by the housing program are "lost" and "moved." The proportion of homeless patients listed on the most recent Case Completion Report as "lost" has decreased steadily from 7.8% in 1996 to 3.3% in 1999. The proportion listed on the most recent Case Completion Report as "moved" dropped from 8.9% among the homeless in 1996 to 4.2% in 1999.

We also analyzed the Case Completion Report (RVCT Follow-up 2) submitted by the jurisdiction that originally reported the case, disregarding subsequent Case Completion Reports submitted by other jurisdictions into which the patient moved. This allows us to determine the outcomes of case management in the jurisdiction that originally reported the case. We were interested in determining if the provision of housing services stabilized patients' living conditions so they would not need to move. This analysis showed that the proportion of homeless patients classified as "moved" decreased from 14.8% in 1996 to 10% in 1998 and to 7% in 1999.

There is evidence to suggest that housing patients had a positive effect on the outcome of therapy. The implementation of this program corresponded with a decreasing proportion of homeless patients whose most recent RVCT 2 indicated that their status was "lost" or "moved." However, since the implementation of the housing program coincided with the implementation of other interventions (e.g., food, increased incentives, enablers, social services, and DOT), it is difficult to precisely determine the effect of the housing program in California. The housing program also coincided with intensified efforts by the CDHS TB Registry to follow up with local health departments to ensure that "moved" was not a final disposition. The latter intervention would not have impacted the proportion of homeless TB patients who move out of the jurisdiction that originally reported them. The trend toward decreasing movement out of the original reporting jurisdiction suggests more successful case management in the original jurisdiction, which may have been aided by the provision of housing.

Housing can be one way to improve TB control while improving patients' lives for at least a short period of time. In addition, while patients are housed, the job of providing DOT and support services is made easier and more cost effective by reducing the time health workers spend searching for patients. However, completion of therapy rates among the homeless remain unacceptably low. The persistent application of many interventions tailored to the individual needs of homeless patients is needed to further improve treatment outcomes.

Submitted by Steve Roger,
Cathryn Fan, and Jan Young
California Department of Health Services Tuberculosis Control Branch

Application of Aggregate Reports to Program Management in Florida

The Aggregate Reports for TB program Evaluation marked a dramatic shift towards improved TB program management and accountability when they were introduced in 1999. Only after completing two cycles of the Contacts Report and one cycle of Targeted Testing did we begin to realize the full potential of these reports.

The reactions of the Florida Department of Health (DOH) TB program field staff were probably much the same as those of staff in other states when the seemingly complicated report formats and new definitions first appeared. They had to learn not just a new vocabulary and new definitions but, in some ways, a completely different way of looking at TB control. This new view of affairs meant that more is not necessarily better; that priorities must be set and results measured. The focus is not on numbers, but on outcomes.

Florida already had in place the nucleus of a TB case management information system before TIMS came along with its case management capabilities. The Florida system, the Health Clinic Management System, or HCMS, contained a registration module, a TB case management and RVCT module, a laboratory module, and limited ad hoc report generation capability. It was logical, if not simple, to base a management reports module for the Aggregate Reports on this platform.

We reached the decision to automate the Aggregate Reports after completing the preliminary contact reports for 1999. With the realization that Florida's 67 county health departments (CHDs) were required to consult treatment records and manual statistical tabulations on approximately 9500 contacts for nearly 1200 cases just for one report, we had to rethink our approach to the preparation of these reports. Through the statewide Quality Improvement process, we also discovered that the old CDC Program Management Reports were being completed and interpreted differently among CHDs. The inescapable conclusion we reached was the necessity of designing a system to standardize and automate the reports.

There is no way to express in simple terms the programming requirements for automating these reports. It is not simply a matter of designing a set of screens that can be used to load the HCMS database to produce the reports for a given cohort year. The committee of users we convened to assist in developing the specifications was forced to develop decision rules applicable to a variety of situations. For example, what cutoffs should be used to count the contacts to cases reported late in the cohort year whose evaluation and treatment took place in the next year?

The initial reports had to be tested, compared to audit reports, and completely validated or returned to the programmer for correction over a period of months, in the course of which the program identified some stellar analytic thinkers. The Contacts Report was completed first, followed in late August 2001 by the Targeted Testing Report. The system was then used to produce the final Contacts Report for 1999, the Preliminary Contacts Report for 2000, and the preliminary Targeted Testing Report for 2000.

If this article were only about the automation of the Aggregate Reports, it could stop here with grateful congratulations to the project managers and programmers who designed the system. But happily the story doesn't stop here.

It is a truism more often expressed than acted upon that data are not collected for their own sake but to serve a purpose. The Aggregate Reports were designed for TB program evaluation as their title indicates, not simply at the national or even the state level but locally as well. Florida has discovered that the aggregate reports can help the CHD TB programs obtain a snapshot of their operating efficiency.

To serve this purpose, we have developed a summary report that compares operating results at the local level to state averages and state goals. This enables each health department TB program to see how it compares to other health departments, whether it is using its resources in the most efficient way, and whether its program is effective in meeting the highest priority needs.

The measures used in the Contacts Report include the following:

  • Other contacts as a percentage of total contacts
  • No contacts rate for smear-positive cases
  • No contacts rate for smear-negative cases
  • Contacts per smear-positive case
  • Contacts per smear-negative case
  • Evaluation rate for contacts to smear-positive cases
  • Evaluation rate for contacts to smear-negative cases
  • Evaluation rate for contacts to "other" cases
  • Disease rates for smear-positive, smear-negative, and "other" contacts (individually)
  • LTBI rates for smear-positive, smear-negative, and "other" contacts (individually)
  • Treatment rates for contacts with LTBI
  • Contacts chose to stop plus contacts lost to follow-up, as a percentage of total contacts who began treatment
  • Completion rates for contacts with LTBI

Certain assumptions underlie these measures. First, a decision to test is a decision to treat. And second, a decision to treat is a decision to complete. Thus, the program should allocate resources to ensure completion of therapy for contacts found to have LTBI.

Line 1, "other" contacts as a percentage of total contacts, is measured to show the percentage of effort given to lower-priority contacts. The amount is not an issue as long as the local TB program has the resources to serve these clients, and its evaluation, treatment, and completion rates for higher-priority contacts meet state standards.

In Lines 2 and 3, the goal for "no contacts" is zero. The state program assumes that every TB case has contacts; and the objective of a contact investigation is to find and evaluate those contacts. There is no goal for number of contacts to "other" cases that presumably have a lower priority.

The goal for lines 4 and 5 is 10 contacts per smear-positive and smear-negative case.

For lines 6-8, Evaluation, the goal is 90%. The whole idea of a contact investigation is to identify and evaluate contacts for possible disease or infection resulting from their exposure to the index case. If the contacts aren't evaluated, the contact investigation is just wasted effort.

There is no goal for lines 9 and 10, TB disease or LTBI, but these measures do serve a beneficial purpose. They show that the contact investigations are achieving tangible results. Also, the LTBI rate may reflect the LTBI background rate in certain population groups. Certainly if the LTBI rate is higher than expected for that community or population group, this may be an indication that the contact investigation net was not spread widely enough, particularly when combined with a low ratio of contacts per case or a high no-contacts rate (or both).

Finally, the goal for lines 11 and 12, Treatment and Completion rates, is 90% for each category. This is the bottom line, both figuratively and literally.

But if line 12 fails to meet expectations, the reason might be found in Line 13, "Contacts who choose to stop or who are lost to follow-up." This rate is compared to a standard of <10%. The rationale underlying this measure is that all contacts receiving treatment for LTBI should be case-managed, and case management should concentrate on persons who show signs that they may not complete treatment. If more than 10% of those who begin treatment fail to complete it, the local program cannot possibly meet the State's 90% completion goal.

While the preceding analysis has focused on identifying weaknesses in the CHD's contact investigation program, the reverse also applies. Strong points are highlighted and documented so that we also know whom to praise for good work as well.

Clearly, it is not sufficient to simply compare health department TB programs one to another or to the state goals. For this process to be useful, it should lend itself to program analysis and management purposes. Also, it should be reproducible at the local level so managers can see how they are progressing periodically. To achieve the second purpose, the program proposes to develop a special HCMS screen that can be accessed locally to display the summary report at any given time for any given time period.

The first priority, program analysis and management, is achieved by understanding what the report shows, and the process to that end seems self-evident. Low contacts per case or high "no-contacts" ratios suggest problems with the contact identification process. This is a critical deficiency, because persons who are not identified cannot be evaluated.

If the evaluation rate is low, it is equally obvious that the deficiency is critical, because persons who are not evaluated cannot be treated. This deficiency may indicate problems in an aspect of the program other than the contact identification process. If this is the problem, the corrective action will differ from that required if the contacts are not being identified. One may be an interviewing deficiency, the other a deficiency in the follow-up procedure to locate and test the contact.

There are also different interpretations that can be given a finding that LTBI rates are too low or too high. The program manager must give some thought to the findings to determine what, if anything, is the significance of the variance from the expected rates.

Persons who have LTBI associated with an infectious TB patient with whom they have contact should be treated. This is the only way to interrupt the transmission of TB. If the treatment rate is low, why is this so and what measures are needed to improve program performance? The same is true for treatment completion. If a person starts treatment, he or she should finish unless there are clinical reasons why completion is not possible. The percentages shown for "contact lost to follow-up" or "contact chose to stop" may provide an explanation as to why treatment completion is low.

It should be evident by now that many of the same measures can be applied to the targeted testing report. Certainly the same principles apply. It should be noted that the Florida DOH chose to implement the targeted testing report beginning with cohort year 2000 for all clients tested or "screened" for LTBI by the county health departments. By doing this, we get summary data of all TB "screening" being done by CHDs across the state and an indication of the degree to which targeted testing guidelines have been implemented throughout Florida. It seems clear that these reports give management a powerful tool with which to analyze their programs' effectiveness. In a time of diminishing resources, programs need every tool possible to deliver measurable results.

Reported by John T. Miller, Bureau of TB
and Refugee Health
Florida Dept of Health


Released October 2008
Centers for Disease Control and Prevention
National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention
Division of Tuberculosis Elimination -

Please send comments/suggestions/requests to:, or to
CDC/Division of Tuberculosis Elimination
Communications, Education, and Behavioral Studies Branch
1600 Clifton Rd., NE - Mailstop E-10, Atlanta, GA 30333