Tuberculosis surveillance using death certificate data, New York City, 1992
Rita M. Washko
Data from death certificates are used to determine disease-specific and overall mortality rates. The major uses for these rates include assessment of the relative public health importance of particular diseases and the efficacy of available treatments.
From 1986 through 1992, tuberculosis (TB) mortality rates in New York City (NYC) remained stable at approximately 3 per 100,000 persons despite a sustained rise in reported TB case rates during this same period (fig.).
Information Ested in the NYC TB registry is verified by NYC Department of Health staff, but death certificate data do not undergo similar review. To determine the accuracy of death certificate TB listings, we reviewed the medical records of all patients whose 1992 death certificates listed TB and reviewed all death certificates of patients known to have died with active TB during 1992.
Accuracy of death certificate TB diagnoses. In general the underlying causes of death listed on death certificates by the certifying physicians are used to determine mortality rates. Although we were primarily interested in TB as the underlying cause of death, we also wanted to evaluate the accuracy of all TB diagnoses listed on death certificates. Therefore, we reviewed 1992 NYC death certificates that listed TB as either the underlying or contributing cause of death.
The International Classification of Diseases, Ninth Revision (ICD-9) codes 010.0 through 018.9 identify TB as the underlying cause of death.1 New York City’s death certificate has a unique coding section for a limited number of outcomes such as TB, human immunodeficiency virus (HIV) infection, and injuries. TB listed as underlying cause of death. receives the appropriate ICD-9 code; if TB is listed elsewhere on the death certificate as a contributing cause, that is noted in the special coding area.
Match with TB registry. We performed computer-assisted queries of the NYC TB registry for all persons with TB listed as the underlying or contributing cause of death on their death certificates; criteria for matching included patient name, age, sex, date of birth, address, and social security number. The reviewer made a determination of the matching outcome for each case. For each,. query, a match required the same name and one of the following: (a) age or date of birth; (b)address; or (c) social security number. Patients with TB confirmed by the registry were considered to have an accurate TB diagnosis. To determine the accuracy of death certificate TB diagnoses, we reviewed the] medical records of those with TB diagnoses on their death certificates but who were not listed in the NYC TB registry.
Individuals are included in the NYC TB registry only if they are confirmed to have positive culture for Mycobacterium tuberculosis or clinical confirmation of active disease. A clinically confirmed case is defined by surveillance case definition for the NYC TB registry as the presence of signs and/or symptoms consistent with TB in an individual who was treated with two or more anti-TB medications.[2,3] Since we evaluated the records of patients who died, the TB diagnosis those not listed in the TB registry was based on bacteriologic or pathologic data; autopsy results were reviewed when available.
The gold standard used to confirm or refute TB disease status was documentation of active TB in case reports or the medical record. Medical records were reviewed for the following information: demographic data; clinical and laboratory information; and TB risk factors such as HIV infection, homelessness, and injectable drug use. Those whose TB treatment was completed before death and who had no clinical, bacteriologic, or histologic evidence of active TB at time of death were considered to have a prior history of TB (inactive). When we found records with incomplete laboratory data, we contacted the laboratories for details. When we did not find sufficient data to substantiate a TB diagnosis, we tried wherever possible to review previous medical records.
Frequency of reporting TB on death certificates. In the second part of the study, we determined how often TB is recorded on death certificates of those who died with active TB. We reviewed copies of the death certificates of people listed in the TB registry as having died with active TB in 1992 to determine if TB was given as either the underlying or contributing cause of death. Active TB was defined as death that occurred less than six months after beginning treatment for TB. Characteristics of patients with active TB at time of death whose TB disease was not recorded on the death certificate were evaluated and compared with those whose TB disease was recorded on the death certificate.
The chi-square test, Fisher’s exact test, or chi-square analysis for linear trend in proportions were used as appropriate.
Accuracy of death certificate TB diagnoses. In 1992 in NYC, 635 death certificates listed TB as either the underlying or contributing cause of death. Of these, 377 (59%) had been reported to the NYC Bureau of TB Control and confirmed as TB cases. Medical records were available for 230 (89%) of the 258 patients who were not confirmed by the TB registry.
Of these 230 patients, 114 (50%) had no evidence of current active TB (most of their medical records mentioned the need to “rule out” TB), and 85 (37%) were infected with nontuberculous mycobacteria (mostly Mycobacterium avium complex). Eighteen (8%) of these 230 had a history of inactive TB (2 to 50 years previously) at time of death; an additional 11 (5%) had a history of TB infection, had no active disease, and were on preventive therapy. Thus, 228 (36%) of 635 persons with TB listed on their death certificates did not have TB. Only two (1%) of the 230 patients with a death certificate TB diagnosis were confirmed as having active TB based on medical record reviews. One of these two was cared for by a private physician who submitted the patient’s specimen to an-but-of-state laboratory; the physician and laboratory failed to report this case of culture-confirmed TB to the NYC Department of Health. In the other instance, TB was diagnosed at autopsy based on the pathologist’s findings and neither the pathologist nor the patient’s physician reported this case to the NYC Department of Health.
TB was listed as the underlying cause of death in 193 of the 635 patients. We were able to confirm the diagnosis of TB for only 129 (67%) of these 193] based on TB registry information and medical record reviews (table 1). The rest did not have TB or the records were incomplete. Patients with TB listed as the underlying cause were more likely to have the diagnosis confirmed than those with TB listed as a contributing cause of death (67% [129/193] versus 57% [250/442]; p = 0.02). The positive predictive value of a listing of TB anywhere on the death certificate was 60% (379/635).[TABULAR DATA 1 OMITTED]
TB Reporting on death certificates. In 1992, 310 people were identified by the TB registry as having had active TB disease at time of death. Of these, 104 (34%) had TB listed on the death certificate as the underlying or contributing cause of death, which included 41 (13%) for whom TB was listed as the underlying cause.
Table 2 lists the most frequent underlying causes of death for those with confirmed TB for whom TB was not listed on the death certificate. Of the 206 patients with TB omitted from the death certificate, 112 (54%) had HIV/AIDS listed as the underlying cause of death. HIV status was known for 214 (69%) of the 310 people who died with TB disease in 1992. The majority of those with a known HIV status were HIV-positive; HIV-positive people were more likely to have TB omitted from their death certificates than those who were HIV-negative (135/206 [66%] versus 2/8 [25%]; RR = 2.6; 95% CI 0.8 to 8.7). However, when HIV-positive persons were compared with persons who either had an unknown HIV status or were Hiv-negative, HIV status was not associated with omission of a TB diagnosis from the death certificate (135/206 [66%] versus 71/104 [68%]; RR = 1.0; 95% CI 0.8 to 1.1) (Table 3). Increasing age was associated with the omission of a TB diagnosis from the death certificate (chi-square for linear trend P = 0.03). The type of hospital did not affect the likelihood of omission of a death certificate TB diagnosis. Of the 310 people who died with TB disease, 67 (22%) died at municipal hospitals; 28 (42%) of these had TB listed on their death certificates. Of the remaining 143, 228 died at private hospitals, 8 died at nonmunicipal long-term care facilities, and 7 died at private residences; 167 (69%) did not have TB listed on their death certificates (p = 0.11). Race, ethnicity, sex, and homelessness were not associated with an omission of TB from the death certificate.[TABULAR DATA 2 OMITTED]
Table 3. Characteristics of 310 patients who died with active TB disease, New York City, 1992
from death certificate
Characteristic Died with active TB(a) Number Percent
0-19 6 3 50
20-44 171 107 63
45-54 60 40 67
55-64 21 17 81[greater than or equal to]65 52 39 75
COPYRIGHT 1996 U.S. Government Printing Office
COPYRIGHT 2004 Gale Group