Tuberculosis among Institutionalized Elderly in Alberta, Canada

Transcription

1 International Journal of Epidemiology International Epidemiological Association 1992 Vol. 21, No. 6 Printed in Great Britain Tuberculosis among Institutionalized Elderly in Alberta, Canada COUN MACARTHUR,* DONALD A ENARSON. #t E ANNE FANNING," f PATRICK A HESSEL* AND STEPHEN NEWMAN* Macarthur C (Division of Pulmonary Medicine, Department of Medicine, University of Alberta, Edmonton, Canada), Enarson D A, Fanning E A, Hessel P A and Newman S. Tuberculosis among institutionalized elderly in Alberta, Canada. International Journal of Epidemiology 1992; 21: Previous studies in the US have suggested that the risk of tuberculosis is increased among eklerty residents of nursing homes. This registry-based study determined and compared the tuberculosis incidence rate for the elderly in nursing homes and community dwellings in Alberta, Canada, over the 5-year period Rate ratios (RR) using the community elderly rate as baseline, were estimated for all notified cases and for culture positive cases only. Adjustment was made for the variables age, sex, and ethnicity. The nursing home elderly in Alberta did not have an increased risk of tuberculosis: adjusted RR = 1.09, 95% Cl : The 20th century has seen a marked shift in the agespecific incidence of tuberculosis disease. At the turn of the century, tuberculosis was primarily a disease of infants and young adults. 1 Nowadays, tuberculosis incidence is highest amongst the elderly, that is, people aged 65 years or over. 2 There are two reasons for this shift in the age-specific incidence. First, the enduring viability of the Mycobacterium tuberculosis organism and second, the steadily diminishing risk of tuberculosis infection over time. Birth cohort studies have convincingly demonstrated that the high rates of tuberculosis disease in old age result from the high rates of infection experienced in earlier life. 3-4 The elderly of today, having lived through an era of endemic, untreated tuberculosis, now constitute the single largest reservoir of infected individuals and, as such, are at risk of active disease through recrudescence of remote, dormant infection. 5 In Canada, during 1981, 28% of all tuberculosis incident cases occurred in those aged 65 years and over Division of Pulmonary Medicine, Department of Medicine, University of Alberta, Edmonton, Canada. * Division of Infectious Disease, Department of Medicine, University of Alberta. t Division of Tuberculosis Services, Department of Health, Government of Alberta. * Department of Health Services Administration and Community Medicine, University of Alberta, Edmonton. Reprint requests: Professor D A Enarson, International Union against Tuberculosis and Lung Disease, Director of Scientific Activities, 68 Boulevard Saint-Michel, Paris, France. an age group that constituted only 9.7% of the total population. 6 Recent studies have suggested that a subgroup of the elderly population those resident in nursing home facilities are at increased risk of tuberculosis disease. 7 " 9 Data from Arkansas in the early 1980s, suggested a two-fold increase in tuberculosis incidence among nursing home elderly compared to community dwelling elderly.' The objective of this registry-based study was to determine and compare the incidence rate of tuberculosis amongst the nursing home and community dwelling elderly in Alberta, Canada, over the same 5-year period, METHODS Population The entire Alberta population 65 years of age and older was selected as the study population. Population figures for the institutionalized and community dwelling elderly were obtained from provincial records and 1981 census data. 10 " All institutional facilities in the province were studied, that is, auxiliary hospitals, nursing homes, special care facilities and senior citizen lodges. (A list of all institutions for the elderly in the province was available from government records.) Only those institutionalized who were aged 65 years or more (81.8"% of the total) were included in the analyses. Nursing home population figures were also adjusted downwards to take into account the average occupancy rate, which in 1981 was 98.4%. An 1175

2 1176 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY assumption was made that the 1981 populations of the two groups of elderly institutionalized and community dwelling remained stable over the 5-year study period. Thus, to calculate person-years contributed to the study, 1981 population figures were multiplied by 5 (Table 1). Measurements Notification records of the Division of Tuberculosis Services, Alberta Health were used to identify tuberculosis cases. Notification of active tuberculosis cases to this registry is mandated by law and a passive casefinding system receives reports from primary care physicians, public health units, microbiology laboratories and pharmacies. (An active case-finding system is in place for follow-up of contacts; people with records of inadequately treated active disease and correctional institution admissions.) Notification of diagnosed cases is believed to be complete, and each reported diagnosis is verified by the Division of Tuberculosis Services. Provincial notification rates, therefore, were used to represent tuberculosis incidence. Data on the site of tuberculosis disease (for example, pulmonary, renal, disseminated) and the method of diagnosis (i.e. culture and/or smear positive and/or clinical impression was abstracted from each case file. The primary independent variable was place of residence at the time of diagnosis. Institutional facilities were identified by a valid residential address. The validity of the residential address provided for each case was determined by comparing three independent sources of information. The residential address on the notification card/case file was compared with that of the contact follow-up list (prepared by a public health nurse) and the home conditions report (prepared by social services). Concordance of these data was greater than 99%. Notified cases from tertiary hospitals were reviewed to determine the date of hospital admission and the residential address prior to admission. Information on potential confounding variables age, sex and ethnic status was also abstracted from the individual case files. The literature has suggested that these variables are associated with both residence in institutional facilities and risk of tuberculosis. Information on two other potential confounders previous tuberculosis disease and concurrent illness was not readily available. With respect to the prevalence of previous tuberculosis disease, survey data from the neighbouring province of British Columbia have indicated that the distribution of this variable is similar for the two groups of elderly studied: i.e. 0.7% and 0.6% of nursing home and community dwelling elderly, respectively, had documented evidence of previous disease. 12 It seemed reasonable to assume a similar distribution of this variable in Alberta, thus no analytic adjustment was made. Analysis Tuberculosis incidence rates for the elderly, by place of residence, were estimated by dividing the number of notified cases in each elderly subgroup by the personyears contributed by that subgroup. Crude rate ratios (RR) by place of residence, were calculated using the incidence rate for the elderly in the community as the baseline. To control for potential confounding, the RR for nursing home elderly was adjusted on the variables age, sex and ethnicity (i.e. native Indian versus 'other'). Native Indian population figures, by age and sex, were not available for the nursing home population. Ethnicity, therefore, was controlled for by restriction, in that Native Indian tuberculosis cases and population figures (numerator and denominator data respectively) were excluded from analyses. The indirect method of standardization was used to adjust for age and sex, with rates for the community dwelling elderly used to calculate expected numbers (Ei). 13 Age was stratified as 65-74, and 85 + years. Stratumspecific rate ratios (adjusted), were estimated by dividing the observed (Oi) by expected (Ei) number of tuberculosis cases. A % 2 test with 5 degrees of freedom, was used to test the homogeneity of the stratumspecific rate ratios. 14 If homogeneity was accepted, a summary adjusted rate ratio 8, was estimated by dividing the sum of the observed (O) by the sum of expected tuberculosis cases (E) over all strata. A x 2 test with 1 degree of freedom was used to test the H o : 9 = 1. The standard error (SE) of 9 was estimated using the equation: SE(9) = O 1/2 /E, and 95% confidence intervals (CI) on 9 were obtained using the equation: 95% CI = 9±1.96 SE(9). 13 The power of the study was estimated using the formula: Z,_p = Z a -2(R 1/2 -l) (E) 1/2, where R is the rate ratio of interest. 15 The smallest detectable rate ratio, given the sample size, at a = 0.05, 1 sided, with 80% power, was The data were analysed in two ways for all notified cases and for culture positive, bacillary cases only. RESULTS The 1981 Census data showed that people or 7.3% of the provincial population were aged 65 years or older. 11 Table 1 provides 1981 population data for

3 TUBERCULOSIS AMONG ELDERLY IN CANADA 1177 TABLE 1 Estimated 1981 population figures for the elderly in Alberta by place of residence TABLE 2 Description of notified elderly tuberculosis cases in A Iberta by place of residence, Institution Number of facilities Population (%) Person-years Nursing home All institutionalized" Community All institutionalized: Nursing homes Auxiliary hospitals Special care facilities Senior citizen lodges Community elderly Overall included * (10%) 6454 (3.9%) 2800 (1.7%) 929 (0.5%) 6138 (3.8%) (90%) Alberta elderly by place of residence. Approximately 10% of the elderly population, people, were institutionalized, of which 6454 people were resident in nursing home facilities. The nursing home elderly thus contributed person-years to the study from 1979 to The institutionalized elderly in total contributed person-years, while the community dwelling elderly contributed person-years (Table 1). Over the 5-year study period, 1074 cases of tuberculosis were notified to the Division of Tuberculosis Services in Alberta. Thus, the overall provincial tuberculosis incidence rate was 9.6/10 5 person-years (p-y). Of the 1074 notified cases, 210 or 19.6%, occurred in people aged 65 years or older giving a tuberculosis incidence rate in the elderly of 25.7/10 5 p-y. Descriptive information on the notified cases in the elderly, by place of residence, is shown in Tables 2 and 3. Of the 210 notified cases in the elderly, 25 cases occurred among institutionalized elderly, with 12 of these in nursing home elderly. As the risk of institutionalization increases with increasing age, not surprisingly, the mean age of the tuberculosis case patients was significantly higher for the institutionalized elderly compared to the community dwelling elderly. Otherwise, no significant differences were found for the 1. Notified cases 2. Mean age (years ± SD) 3. Sex (female) 4. Ethnicity (Native Indian) 5. Previous disease ± 8.6 4(33) 3 (25) 2 (17) 9 (75) * Includes nursing home cases ± (48) 3 (12) 4 (16) 20 (80) ± (57) 48 (26) 28 (15) TABLE 3 Site of diagnosis and place of residence of notified eideriy tuberculosis cases in Alberta All Site of diagnosis Nursing home institutionalized* Community No. (Vt) Pulmonary Other respiratory Miliary Adenitis Genitourinary Abdominal Skeletal Other Total 10 (83.3) 1 (8.3) 1 (8.3) 12 (100) * Includes nursing home cases. TABLE 4 Crude incidence rates for tuberculosis in the elderly by place of residence in Alberta (80) 2 (8) 25 (100) 140 (76) 15 (8) 8 (4) 9(5) 6(3) 3 (2) 1 (1) 3 (2) 185 (100) distribution of the variables sex, ethnicity and previous disease among the cases in the two elderly groups. Approximately 80% of the notified cases in the institutionalized and community elderly groups had culture positive proof of tuberculosis disease. Incidence rates for the elderly by place of residence are shown in Table 4, together with unadjusted RR and 95% CI. The Place of residence Number of cases Rate/ p-y RR 95% CI Community Institutionalized Nursing home Auxiliary hospital Special care facilities Senior citizen lodges ( ) ( ) ( ) ( ) ( )

4 1178 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY largest crude RR was noted for the nursing home elderly (RR = 1.48); however, all CI included unity. The RR for the nursing home elderly compared to the community dwelling elderly, adjusted for the variables age, sex, and ethnicity are shown in Table 5. Compared to the crude estimate, adjustment on age lowered the RR, whereas adjustment for sex led to an increased RR. Adjusting on the variables age, sex and ethnicity simultaneously, resulted in a RR smaller than the crude estimate and also not significantly different from unity (RR = 1.09, 95% CI: ). The adjusted RR for nursing home versus community elderly for culture positive cases only, was also not signficant: RR = 1.22, 95% CI : TABLE 5 RR for the nursing home elderly versus community elderly adjusted for age, sex, and ethnicity Variable Summary RR CI Age Sex Age, sex and ethnicity ( ) ( ) ( ) DISCUSSION The finding of this study that institutionalized elderly in general, and nursing home elderly in particular are not at increased risk of tuberculosis is in contrast to previous studies. 7 " 9 Methodological differences in study design and/or demographic differences in study populations may explain this discrepancy. A methodological difference between the current and previous studies is the potential for surveillance bias in the previous studies. In these studies, nursing home elderly were screened and followed over time for the development of tuberculosis. Diagnostic tools utilized included repeat tuberculin skin testing; chest X-rays; sputum culture and clinical impression. The incidence rate of tuberculosis among nursing home elderly was then compared with the incidence rate amongst the community dwelling elderly. However, the community dwelling elderly were not followed over time, that is, the incidence rate in this group was based on data provided by voluntary physician reporting to the State Health Departments. Low reporting rates for tuberculosis have been associated with the passive reporting process used to identify tuberculosis cases among the community elderly. 16 Therefore, the RR for tuberculosis risk, determined by comparing nursing home elderly with community elderly in these studies, may have been overestimated. Lastly, only crude RR were used: no design or analytic adjustment was made for potential confounders such as age, sex, race, previous disease or concurrent illness. Demographic differences between the study cohorts may also account for the discrepancy in RR estimates. Alberta is a 'young' province with only 7.3% of the population aged 65 years or over, and less than 1% aged 85 years or over. In addition, the ethnic mix of elderly is predominantly Caucasian and the standard of living is relatively high. As a result the tuberculosis incidence rate in the elderly in Alberta is low 25.7/ per year. This rate is approximately 2.5 times lower than that for the Arkansas elderly. 17 It may be argued that epidemic spread of tuberculosis in nursing homes requires a higher background prevalence than that of Alberta. In addition, the nursing home elderly population in the US (compared to the Alberta nursing home elderly) may be, 'on average', older, less healthy and of an ethnic mix for whom tuberculosis is more prevalent compared to the community elderly. If so (without adjustment) this distribution of risk factors would lead to an increased RR for tuberculosis among US nursing home elderly compared to the community elderly. The role played by differences in nursing homes care between Alberta and the US which may result in variation in tuberculosis risk, e.g. factors such as overcrowding or inadequate ventilation, are more difficult to address, given the dearth of pertinent data. This study found no significant difference in the tuberculosis incidence rate between the nursing home and community dwelling elderly in Alberta. Sample selection bias was considered unlikely given that the entire Alberta elderly population was chosen for study. The possibility of differential misclassification of exposure or outcome status was also considered unlikely. The exposure of interest, that is, the residential address at the time of diagnosis, was validated using three independent sources. Concordance of these data was greater than 99%. In addition, confidence in the validity of the residential address, particularly for the institutionalized elderly, was high for a practical reason. That is, extensive follow-up contract lists consisting of nursing home residents and auxiliary staff, were 'standard' for all nursing home cases identified, thus easing identification. The issue of misclassification of outcome was addressed by considering culture positive, bacillary, cases only. In developed countries, given the technology available, approximately 80% of tuberculosis cases should be culture proven.'overdiagnosis of 'cases', e.g. as a result of active screening, will decrease the proportion of bacillary cases. Surveillance bias was considered unlikely in this study

5 TUBERCULOSIS AMONG ELDERLY IN CANADA 1179 given the similar proportion of bacillary cases in the two groups of elderly both close to 80%. The issue of underdiagnosis of tuberculosis cases is more difficult to assess. It may be argued that the likelihood of tuberculosis diagnosis is increased in the nursing home setting. That is, many nursing home residents in Alberta are screened for tuberculosis on admission to the facility and regularly examined thereafter. If this assumption is true, then the RR in this study would have been overestimated. Adjustment for age, sex and ethnicity was undertaken in our study. Adjustment on the variable, previous disease was not possible and is a shortcoming of the study. However, even if it is assumed that people with previous disease were more likely to enter nursing homes because of compromised lung function, the direction of the bias would be to overestimate the RR. Similarly, for the variable concurrent illness, using the same argument, the direction of the bias would again be to overestimate the RR. With respect to the powet of the study, a priori, an important difference in the tuberculosis incidence rate between the nursing home and the community elderly was declared to be RR of 4 or more. This magnitude of difference was chosen for both academic and practical reasons. First, the Arkansas studies suggested a fourfold difference in incidence rates of tuberculosis between the two groups of elderly. 9 In addition, a common rule of thumb in tuberculosis surveillance is that screening is cost-efficient when the disease prevalence is greater than 1/1000 people. Therefore, given a background rate of 25.7/ in the Alberta elderly, a four-fold increase in tuberculosis incidence amongst the nursing home elderly would satisfy this screening criterion. This study had greater than 99% power to detect an RR of 4 or more. A 10-year retrospective survey in the province of Saskatchewan, Canada also reported no increase in tuberculosis incidence among the elderly in chronic care homes. 18 The issue, at least in Canada, is whether screening for tuberculosis in the nursing home elderly is justified in terms of cost-benefit ratio. Both provincial and national regulations recommend screening for tuberculosis among the institutionalized elderly. 19 ' 20 Tuberculosis surveillance requires several diagnostic tools repeat tuberculin tests, chest X-rays and sputum cultures. In this study, over 5 years, only one case of tuberculosis in the nursing home elderly was identified by screening at the time of admission to the facility. To continue the practice of routine screening, given the low prevalence in the nursing home population, seems particularly cost inefficient in this era of cost-containment. 21 ACKNOWLEDGEMENTS Dr Macarthur was a Province of Alberta Graduate Scholar and is presently a National Health Research and Development Program Doctoral Fellow. REFERENCES Springett V H. An interpretation of statistical trends in tuberculosis. Lancet 1952; I: Powell K E, Farer L S. The rising age of the tuberculosis patient: A sign of success and failure. / Infect Dis 1980; 142: 946-^18. Frost W H. The age selection of mortality from tuberculosis in successive decades. Am J Hyg 1939; 30: Grzybowski S, Mar W B. The unchanging pattern of pulmonary tuberculosis. Can Med Assoc J 1963; 89: 737^*0. 5 Davidson P T. Tuberculosis: New views of an old disease. NEnglJ Med 1985; 312: Tuberculosis Statistics, Morbidity and Mortality, Statistics Canada Catalogue Stead W W. Tuberculosis among elderly persons: An outbreak in a nursing home. Ann Intern Med 1981; 94: Narain J P, Lofgren J P, Warren E, Stead W W. Epidemic tuberculosis in a nursing home: A retrospective cohort study. J Am Genatr Soc 1985; 33: Stead W W, Lofgren J P, Warren E, Thomas C. Tuberculosis as an epidemic and nosocomial infection among the elderly in nursing homes. N Engl J Med 1985; 312: Alberta Hospitals and Medical Care. Annual Report 1980/ Census of Canada: Population. Statistics Canada Catalogue Grzybowiki S, Allen E A, Chao C W. Screening for tuberculosis in elderly nursing home residents. In: Mycobacteria of Clinical Interest. Amsterdam: EUevier Science Publishers, 1986, pp Breslow N E, Day N E. Statistical Methods in Cancer research. Volume II. The design and analysis of cohort studies. LARC Scientific Publications, 1987; pp Armitage P, Berry G, Statistical Methods in Medical Research (2nd edn). London: Blackwell Scientific Publications, 1987, pp Beaumont J J, Breslow N E. Power considerations in epidemiologk studies of vinyl chloride workers. Am J Epidemiol 1981; 114: Marier R. The reporting of communicable diseases. Am J Epidemiol 1977; 105: Stead W W. Tuberculosis in the elderly (letter). J Am Geriatr Soc 1983; 32: Hoeppner V H, Ring E D. Tuberculosis in chronic care homes. Can Fam Physician 1987; 33: Tuberculosis control in the Province of Alberta. Edmonton, Alberta: Tuberculosis services, Canadian Lung Association. Canadian Tuberculosis Standards. Toronto: University of Toronto Press, Enarson D A, Wade J P, Embree V. Risk of tuberculosis in Canada: Implications for priorities in programs directed at specific groups. Can J Public Health 1987; 781: (Revised version received June 1992)