Abstract
Although asthma is on the rise in the United States and elsewhere, data on age–sex-specific patterns of change in various types of health care utilization are scarce. We report on 20-yr trends in the treated prevalence of asthma among members of a large health maintenance organization. Data are presented separately for each of six age–sex categories, and include both the treated prevalence of asthma as well as the treated prevalence of the broader category of chronic airflow obstruction (CAO), defined as asthma, chronic bronchitis, or emphysema. During the period 1967–1987 the treated prevalence of asthma and CAO increased significantly in all age–sex categories except males aged 65 and older. These patterns are in contrast to previous studies of this population that showed that increases in asthma hospitalizations and hospital-based episodes of care were limited primarily to young boys. Not only do these findings support other evidence of a real increase in asthma prevalence, but they also highlight the risks associated with drawing inferences about changing disease epidemiology based on a single type of health care utilization.
Although the reasons are still not fully understood, asthma has been on the rise in the United States and elsewhere since the mid-to-late 1970s. Early evidence for an increase in asthma morbidity came from analyses of hospital discharges and, to a lesser extent, mortality statistics (1-4). Although national data from the United States show increases in asthma hospitalizations among older as well as younger individuals, (2, 5), these increases have been most pronounced in children, and especially in children under the age of five (5-8). Studies of emergency department use subsequently confirmed these findings (8).
Studies of trends in asthma prevalence in the United States, though less numerous, support these trends. Much of what we know about trends in asthma prevalence derives from large, government-sponsored surveys that provide data at selected points in time. For example, U.S. data from 1982 through 1992 demonstrate an increase in the overall age-adjusted prevalence rate of self-reported asthma from 34.7 per 1,000 to 49.4 per 1,000 (42% increase) (10). Increases in asthma prevalence have also been reported for England, Canada, Israel, and Australia (11-14).
Although these studies paint a convincing picture of increasing asthma prevalence, morbidity, and mortality, the full nature and extent of these increases, much less their causes, are still not fully understood (15). For example, at least part of the increase in asthma hospitalizations and emergency department visits appears to result from corresponding decreases in chronic bronchitis (3, 5, 6, 9). Similar patterns may hold for prevalence estimates. Data from the National Morbidity Study in England and Wales, for instance, show not only a sharp increase in the rate of asthma-related visits to a general practitioner between surveys conducted in 1970 and 1980, but also a concomitant drop in visits for chronic bronchitis (16). More recent data shows this trend continuing into the 1990s, with children exhibiting the greatest increases in asthma visits (17).
In general, differences in age–sex patterns for asthma prevalence trends have not been well studied, and it is not clear if patterns of change in asthma prevalence mimic those for asthma hospitalizations and emergency department use. For example, are increases in asthma prevalence also most pronounced among young boys? Such questions can perhaps best be addressed by examining different utilization statistics in the same population over time.
This report describes 20-yr trends in the treated prevalence of asthma among members of a large health maintenance organization (HMO). Data are presented for six age–sex subgroups and are shown for consecutive 3-yr time intervals. We also present trends for chronic bronchitis and emphysema over this same time frame, and for all three disorders combined. We compare our findings with those obtained in previous analyses of trends in hospital admissions and emergency department care in this same population (6, 9). Collectively, this series of studies represents the first systematic attempt to study trends in a variety of health care utilization outcomes in a single, well-defined population.
Kaiser Permanente (KP), Northwest Division, is a large, federally qualified, group model HMO that provides comprehensive, prepaid health care service to its members. The service area is centered in Portland, Oregon, and extends in a radius of about 50 miles. During the years covered by this study (1966–1987), KP enrollment grew from 86,200 to 310,800. The demographic and socioeconomic characteristics of KP members corresponded to those of the area population as a whole during this period (18, 19), and no evidence exists of systematic selection of healthy individuals either into or out of the system (20, 21).
Data for this study are derived primarily from the abstracted outpatient medical records of a random sample of KP members and from an eligibility file that tracks health plan eligibility for members included in the outpatient utilization sample. In Figure 1 we also include a limited amount of hospital discharge data. The hospital discharge data were coded using the International Classification of Disease (ICD) classification system in force at the time of discharge, and was available for all health plan members. We briefly describe the outpatient database below. More detailed descriptions of this database and of the hospital discharge database may be found elsewhere (6, 9, 18).
Fig. 1. Comparison of utilization trends for asthma among women aged 65 and older in Kaiser Permanente, Northwest Division, 1967–1987. Data for treated prevalence plotted against the right axis and data for episode and hospitalization rates plotted against the left axis. Treated prevalence and episode data coded consistently throughout (see Methods). Hospitalization data coded using ICD system in effect at time of discharge (ICD-7 for 1966–1969, ICD-8 for 1970–1976, ICD-9 for 1979–1987, no data available for 1977–1978).
[More] [Minimize]The outpatient utilization system contains the abstracted records of a simple random sample of KP subscriber units (typically families). Prior to 1984, 5% of KP subscriber units were included in this sample. Beginning in January 1984 and continuing through 1991, the outpatient database contained the abstracted medical records of 2% random sample of subscriber units. Chart abstraction for the 2% sample was discontinued in 1992. The data set upon which the present analysis is based was established in 1989 as part of an NIH-funded grant looking at trends in the clinical epidemiology of asthma within the health plan from 1967–1987. This represents the third in a series of articles (6, 9) looking at patterns in health care utilization over that time frame.
As part of the chart abstraction process, all morbidities (conditions) for which a patient received treatment were recorded for each visit using a system, described below, that was specifically designed to deal with the problem of coding ambulatory care data (22). For each contact with the health care system, a presenting morbidity was determined as the primary reason for the visit as listed in the chart. In addition, up to eight associated morbidities could be listed. All morbidities carry an updated diagnosis that could be modified as new information was recorded in the medical record. For example, an initial diagnosis of nighttime cough might be retrospectively updated to asthma if the physician subsequently decided that this was a manifestation of previously undiagnosed asthma. We used only updated diagnoses for greater accuracy. The outpatient database also records primary and secondary discharge diagnoses for subjects who are hospitalized.
An eligibility file provided a detailed monthly eligibility history for every member included in the outpatient database. This permitted us to associate person-years of eligibility with our outpatient utilization data. Table 1 presents the total person-years of eligibility for the various age–sex–time categories used in the analysis. These figures represent the denominator population for all of the treated prevalence data.
The KP disease classification system codes for asthma correspond to the ICDA-7 codes (241.0–241.9). Because asthma may be confused with other manifestations of airflow obstructive disease, we also examined utilization data for chronic bronchitis and emphysema (ICDA-7 codes 502.0, 502.9, and 527.1).
Although we ostensibly used the same coding system over time, we did periodically adapt our codes in order to incorporate some of the revisions to the ICD system that occurred over time. Of particular relevance to this study, the term “asthmatic bronchitis” was initially coded under ICDA-7 code 501.0 (bronchitis, unspecified), which we stopped using in 1970. Diagnoses falling under this code were subsequently coded under ICDA-7 codes 500.0–500.9. Since these latter codes include primarily acute bronchitis, and since the former 501 codes are now intermingled with these codes, we chose to include neither the 500 nor 501 codes in our definitions of asthma, bronchitis, or emphysema. Consequently, subjects classified as having asthmatic bronchitis are not included in this analysis.
The data are presented in 3-yr time intervals for greater stability and are expressed as weighted averages of the separate annual treated prevalence estimates (i.e., average annual treated prevalence). For any given year and age–sex category, the annual treated prevalence was computed as the number of person-months of eligibility for those with asthma visits (either as a presenting or associated morbidity) during that year divided by the total person-months of eligibility for people in that age–sex category for that year. This formula takes into account the known variation in eligibility from person to person, and reverts to the more common prevalence formula (proportion of eligible members treated during the year) when the number of person-months of observation is the same for all members. The age intervals used in this analysis were chosen to provide comparability with our previous analyses of these data (6, 9).
For data analysis we assumed that each average annual treated prevalence estimate could be approximated as a binomial random variable with sample size given either as the total person-years of eligibility over the 3-yr period or as one-third of this figure. The former number is the most liberal estimate and assumes complete independence of the three annual treated prevalence estimates. The latter sample size estimate, by contrast, is the most conservative figure. Together the two sample size estimates provide a confidence band for performing inference. We also assumed independence of average annual treated prevalence estimates across age–sex–time categories. Under these assumptions the data can be analyzed using logistic regression, with differences between competing models compared using likelihood ratio methods (23). While we recognize the independence assumption is not totally valid, we feel it does provide a reasonable basis for inference.
In addition to treated prevalence data, Figure 1 also provides data on the rate of hospital discharges for asthma and on the rate of hospital-based episodes of asthma care. The methodology for defining these latter statistics has been described elsewhere (6, 9). Although we feel that the conclusions drawn from Figure 1 are valid, we should note that, while the episode data were constructed from the same outpatient data set as used here, the hospital discharge data were derived from a separate data set that included all hospitalizations and coded diagnoses according to the ICD system in effect at the time of discharge. As a result, hospitalizations for “asthmatic bronchitis” are included in the hospital discharge rates shown in Figure 1 from 1980 through 1987. Prior to that time they were coded under the rubric of “chronic bronchitis” and hence are not reflected in the figure.
Unless otherwise stated, all p values are two-sided and the term “significant” implies a p value less than or equal to 0.05.
Figure 2 shows the trends in treated prevalence for asthma for males and females in each of three age groups: 0–14-yr-olds, 15–64-yr-olds, and 65+. The figure represents the proportion of members seen annually for asthma, whether as a presenting or associated morbidity. The figures are highest in the youngest and oldest age groups and, with the exception of men over age 65, show a clear pattern of increase over time for each age–sex group. Using both our conservative and more liberal estimates of sample size (see Methods), the treated prevalences differed significantly over time after adjusting for age and sex. With the more conservative estimate, these differences were independent of age and sex, although using the larger sample size estimate, the time effects did differ significantly be age and sex. This latter interaction was limited to males and is consistent with the lack of any secular increase in older males in contrast to the increases seen in the two other age groups.
Fig. 2. Estimated annual treated prevalence of asthma (%) in Kaiser Permanente, Northwest Division, 1967–1987.
[More] [Minimize]Particularly for the older cohort, the prevalence patterns for asthma may be affected by changing patterns of diagnosis for chronic bronchitis and emphysema. Table 2 presents, for the over 65 group, the treated prevalence data for the combined diagnosis of either chronic bronchitis or emphysema, and also for asthma, chronic bronchitis and emphysema combined (chronic airflow obstruction). The treated prevalence of chronic bronchitis/emphysema was much more common among older men than among older women throughout this time period and, among the men, remained relatively flat. Among older women, by contrast, the average annual treated prevalence of chronic bronchitis/emphysema rose steadily from 0.9% per year to 4.2% per year over the 20 yr. While some of the increase in asthma prevalence in this group may have resulted from a diagnostic shift from chronic bronchitis/emphysema to asthma over this time frame, these data make clear that the overall burden of chronic airflow obstruction in older women is truly on the rise and is in marked contrast to the pattern seen for older men.
| Year | Chronic Bronchitis/ Emphysema | Chronic Airflow Obstruction† | ||||||
|---|---|---|---|---|---|---|---|---|
| Male | Female | Male | Female | |||||
| 1967–69 | 8.9 | 0.9 | 10.5 | 1.5 | ||||
| 1970–72 | 9.8 | 0.7 | 11.0 | 1.6 | ||||
| 1973–75 | 11.7 | 1.3 | 13.1 | 2.0 | ||||
| 1976–78 | 8.6 | 1.8 | 9.5 | 3.0 | ||||
| 1979–81 | 7.4 | 1.5 | 8.2 | 3.4 | ||||
| 1982–84 | 8.5 | 2.5 | 9.6 | 4.9 | ||||
| 1985–87 | 9.9 | 4.2 | 10.7 | 6.1 | ||||
Figure 3 presents the temporal trends for the combined chronic airflow obstruction category for all three age groups. Note that for the older cohort the scale is compressed to accommodate the much higher treated prevalence in males. The data in these three panels are obviously weighted predominantly toward asthma in the youngest cohort and, especially for the males, predominantly toward chronic bronchitis/emphysema in the over 65 cohort. As with the asthma-specific data, the trends increase steadily over time in all but males in the oldest age category. Logistic regression analysis again confirmed the significance of these trends. Furthermore, even for the conservative sample size estimate, the time trends among males differed significantly across the three age groups.
Fig. 3. Estimated annual treated prevalence of chronic airflow obstruction (%) in Kaiser Permanente, Northwest Division, 1967– 1987. Note the different scale for the 65 and older group.
[More] [Minimize]Previous analyses of this population have examined trends in both the rate of hospital discharges for asthma and in the rate of occurrence of hospital-based care, including emergency department and urgency care service (6, 9). The results of these previous analyses suggested very different pictures of how asthma has changed over time in this population. This contrast is illustrated in Figure 1 for women over age 65. The figure superimposes asthma utilization statistics for three different outcomes: hospital admission rates, rate of hospital-based episodes of care, and treated prevalence. The increase in hospital admissions in 1979 reflects the shift from ICD-8 to ICD-9, and the consequent inclusion of asthmatic bronchitis under the category of asthma. As noted in Methods, asthmatic bronchitis is excluded from the prevalence and episode calculations. Other than for the ICD-induced shift, the hospital-based data do not exhibit a consistent pattern of increase over time, whereas the treated prevalence data show a striking increase over time.
Three main findings emerge from this study. First, for the 20-yr period from 1966–1987 the treated prevalence of asthma increased steadily and significantly in this population in both males and females and in all age ranges except males over 65 yr of age. Second, these increases parallel increases in the broader category of chronic airflow obstruction and therefore are not likely merely to reflect diagnostic shift from chronic bronchitis/emphysema toward asthma in the midst of an otherwise stable pattern of chronic airways disease. In fact, the prevalence of both asthma and chronic bronchitis/emphysema appear to be increasing steadily in older women. These results also demonstrate graphically the dangers of extrapolating trends in one type of asthma health care utilization outcome to other types of health care utilization outcomes.
The weight of evidence that asthma is on the rise, both in terms of prevalence and health care utilization, is growing steadily, even allowing for the likelihood that some of the reported increase in morbidity and mortality reflects diagnostic shift or increasing case-finding on the part of physicians (10– 14, 16, 17). Although the time frame encompassed by the present study, 1967–1987, does not provide insights into more current trends in asthma prevalence and health care utilization, more recent national data suggest that these increases are continuing (10). The data presented here do support the conclusion that asthma is on the rise and also provide new information on the extent of the problem, both in terms of age and gender patterns. Moreover, these results, when coupled with our previous analyses of this population (6, 9), represent the most detailed analysis currently available of trends in health care utilization for asthma and, more generally, chronic airflow obstruction, in a single reference population.
The generalizability of our results may be limited by the nature of the population. The members of a large HMO obviously do not constitute a population-based sample in the usual sense. Nonetheless, KP is a mature health plan that, even in 1966, had considerable market penetrance. The available data suggest that, over the time frame encompassed by the study, the demographic and socioeconomic characteristics of KP members correspond well to those of the area population as a whole, although the membership does underrepresent the extremes of wealth (18, 19). Further, no evidence exists of a systematic selection of healthy individuals either into or out of the system, although the same may not be true for other health maintenance organizations (20, 21). Also, the expansion of Medicare coverage in 1980 to include capitated health care resulted in a substantial growth of members aged 65 and older, and it is important to look at this age group separately when comparing trends in health care utilization over time.
Another factor that may limit the generalizability of our results is the exclusion of visits for asthmatic bronchitis from the analysis. As noted in Methods, midway through the study period coding of this term shifted from the rubric “bronchitis, unspecified” to what is essentially “acute bronchitis.” In either case it would not have been classified as asthma, although initially it would have been included under our broader category of chronic airflow obstruction. In order to avoid an artificial drop in the treated prevalence of the latter category, we opted to exclude the “bronchitis, unspecified” code totally from our analysis. The net result of this decision is that the true treated prevalence of asthma may be underestimated by our data, but we believe the temporal trends in treated prevalence should be largely unaffected.
Numerous authors have speculated on possible reasons for the increase in asthma prevalence, morbidity, and mortality (15, 24-27). Among the factors that have been suggested as possible explanations are increasing concentrations of indoor airway allergens, changing patterns in the immune response to infections, increased consumption of specific dietary nutrients, changes in the organization and delivery of health care, and the possibility of adverse drug effects.
Similarly, gender differences in the epidemiology of asthma have long been recognized, especially in children (28), although gender-related differences in trends of asthma care are less well studied. Gender differences in asthma, particularly with aging, have been ascribed to hormonal status, relative differences in airway and parenchymal size, differences in environmental exposures, and even differences in asthma management (29). Also, Goodman and colleagues (30) have demonstrated that women of all ages are more likely than men to exhibit improper metered-dose inhaler technique.
The most striking gender-related differences in our study were in the trends seen for the treated prevalence of asthma and overall chronic airflow obstruction in the 65 and older cohort. While chronic airflow obstruction remained consistently more common in men than in women, the pattern remained relatively flat for men but has been steadily increasing among women. For asthma, the pattern was similar except that the overall treated prevalence among women caught up to, and eventually surpassed, that among men during the period of study. These findings generally corroborate gender differences in overall chronic obstructive pulmonary disease (COPD) mortality for the United States during the same time frame, which increased steadily for women aged 45 and older while remaining relatively flat (although still higher than for women) among similarly aged men (31). Interestingly, the overall increase in COPD mortality reported in that study, which mirrors the overall pattern of increased prevalence of chronic airflow obstruction seen here, was in marked contrast to declining mortality from cardiovascular causes and all other causes.
These gender differences may be related to trends in cigarette smoking for men and women. Although overall cigarette smoking has been declining in the United States since the first Surgeon General's report was released in 1964 (32), most of this decline occurred among men, while the prevalence of smoking among women was either holding steady or increasing between 1955 and 1975 (33). Additional differences may result from diagnostic biases on the part of physicians. Dodge and colleagues (34) reported that, given the same presenting symptoms and controlling for smoking, women in Tucson were more likely to be diagnosed as having asthma while men were more likely to be diagnosed as having emphysema.
Perhaps the most intriguing and important aspect of these results is the contrast between the patterns we observed for the treated prevalence of asthma and those we observed for emergency department use (9) and hospital admissions (6) in this same population. We found that the increase in hospital- based care for asthma, and indeed for chronic airflow obstruction generally, was restricted primarily to young boys. Further, these findings were similar to those of many other researchers (5, 7, 8). The present study, by contrast, paints a much different picture of the growing impact of asthma and other obstructive airways disorders in this population. The increases we observed in treated prevalence were present in both males and females and across all ages. Only in older men did we observe no change.
This contrast in findings among the three studies points out the difficulties inherent in extrapolating trends in one type of asthma outcome to other types of outcomes, as we have previously discussed (35). For example, asthma hospitalizations, while accounting for a disproportionate part of total asthma health care costs, reflect only the tip of the iceberg in terms of total asthma care visits. Further, trends in hospitalization data can be confounded by diagnostic shift among physicians, changes in access to and financing of medical care, changing patterns of emergency asthma management, and changing hospital admission criteria. Emergency department visits, though more common than hospitalizations, are still relatively uncommon events and subject to many of the same limitations as are hospitalization statistics. Furthermore, in the managed care environment the concept of emergency or urgent care is becoming increasingly elusive in terms of both the physical provision of care (e.g., same-day care, walk-in care, after-hours care, etc.) and the way visits are captured electronically (9).
Historically, very little information has been published about patterns of general outpatient care (35). For the most part, this has probably reflected the lack of computerized databases containing diagnosis-specific outpatient data. Such databases are becoming increasingly common, however, and the information they provide should open important new areas of research in both epidemiology and health services research.
The authors thank the staff of the Kaiser Permanente Research Medical Information Department, Reesa Laws, and Barbara Mendius for their assistance in the creation and analysis of this data set.
Supported by Public Health Service grant no. HL41039 from the National Heart, Lung, and Blood Institute, NIH.
| 1. | Sly R. M.Increases in deaths from asthma. Ann. Allergy5319842025 |
| 2. | Evans R., Mullally D. I., Wilson R. W., Gergen P. F., Rosenberg H. M., Grauman J. S., Chevarley F. M., Feinleib M.National trends in the morbidity and mortality of asthma in the US: prevalence, hospitalization and death from asthma over two decades: 1976–1984. Chest91198765S74S |
| 3. | Halfon N., Newacheck P. W.Trends in the hospitalization for acute childhood asthma, 1970–84. Am. J. Public Health76198613081311 |
| 4. | Jackson R., Sears M. R., Beaglehole R., Rea H. H.International trends in asthma mortality: 1970 to 1985. Chest941988914919 |
| 5. | Gergen P. J., Weiss K. B.Changing patterns of asthma hospitalization among children: 1979 to 1987. J.A.M.A.264199016881692 |
| 6. | Vollmer W. M., Buist A. S., Osborne M. L.Twenty-year trends in hospital discharges for asthma among members of a health maintenance organization. J. Clin. Epidemiol.4519929991006 |
| 7. | Anderson H. R., Bailey P., West S.Trends in hospital care of acute childhood asthma. B.M.J28119801191 |
| 8. | Mitchell E. A., Cutler O. R.Pediatric admissions to Auckland hospital for asthma from 1970–1980. N.Z. Med. J.961983727729 |
| 9. | Vollmer W. M., Osborne M. L., Buist A. S.Temporal trends in hospital-based episodes of asthma care in a health maintenance organization. Am. Rev. Respir. Dis.1471993347353 |
| 10. | 1995. Asthma: United States, 1982–1992. M.M.W.R. 43:952–955. |
| 11. | Anderson H. R., Butland B. K., Strachan D. P.Trends in prevalence and severity of childhood asthma. B.M.J.308199416001604 |
| 12. | Erzen D., Roos L. L., Manfreda J., Anthonisen N. R.Changes in asthma severity in Manitoba. Chest10819951620 |
| 13. | Kivity S., Shochat Z., Bressler R., Wiener M., Lerman Y.The characteristics of bronchial asthma among a young adult population. Chest10819952427 |
| 14. | Peat J. K., van den Berg R. H., Green W. J., Mellis C. M., Leeder S. R., Woolcock A. J.Changing prevalence of asthma in Australian children. B.M.J.308199415841585 |
| 15. | Buist A. S., Vollmer W. M.Reflections on the rise in asthma morbidity and mortality. J.A.M.A.264199017191720 |
| 16. | Department of Clinical Epidemiology, National Heart and Lung Institute, LondonRespiratory disease in England and Wales. Thorax431988949954 |
| 17. | Lung and Asthma Information Agency, Department of Public Health Sciences, St. George's Hospital Medical School, London. 1996. Respiratory Morbidity in General Practice, 1971–1991. Factsheet 96/3. |
| 18. | Greenlick M. R., Hurtado A. V., Pope C. R., Saward E. U., Yoshioka S. S.Determinants of medical care utilization. Health Serv. Res.31968296315 |
| 19. | Greenlick, M. R., D. K. Freeborn, and C. R. Pope. 1988. Health Care Research in an HMO: Two Decades of Discovery. Johns Hopkins University Press, Baltimore. |
| 20. | Freeborn D., Pope C.Health status, utilization and satisfaction among enrollees in three types of private health insurance plans. Group Health J.31982411 |
| 21. | McFarland B. H., Freeborn D. K., Mullooly J. P., Pope C. R.Utilization patterns and mortality of HMO enrollees. Med. Care241986200208 |
| 22. | Hurtado A. V., Greenlick M. R.A disease classification system for analysis of medical care utilization, with a note on symptom classification. Health Serv. Res.61971235250 |
| 23. | McCullagh, P., and J. A. Neider. 1983. Generalized Linear Models. Chapman and Hall, New York. |
| 24. | Sears, M. R. 1986. Why are deaths from asthma increasing? Eur. J. Respir. Dis. 69(Suppl. 147):175–181. |
| 25. | Burney P.Asthma deaths in England and Wales, 1931–1985: evidence for a true increase in asthma mortality. J. Epidemiol. Commun. Health421988316320 |
| 26. | Anderson H. R.Is the prevalence of asthma changing? Arch. Dis. Child641989172175 |
| 27. | Hatch G. E.Asthma, inhaled oxidant, and dietary antioxidants. Am. J. Clin. Nutr.611995625S630S |
| 28. | Weiss S. T., Gold D. R.Gender differences in asthma [editorial]. Pediatr. Pulmonol.191995153155 |
| 29. | Redline S., Gold D.Challenges in interpreting gender differences in asthma. Am. J. Respir. Crit. Care Med.150199412191221 |
| 30. | Goodman D. E., Israel E., Rosenberg M., Johnston R., Weiss S. T., Drazen J. M.The influence of age, diagnosis, and gender on proper use of metered-dose inhalers. Am. J. Respir. Crit. Care Med.150199412561261 |
| 31. | Higgins, M. W., and T. Thom. 1989. Incidence, Prevalence and Mortality: Intra- and intercounty differences, Chap. 2. Marcel Dekker, Inc., New York. |
| 32. | U.S. Surgeon General. 1964. Smoking and Health. U.S. Department of Health, Education, and Welfare, Washington, D.C. |
| 33. | Bartecchi, C. E., T. D. MacKenzie, and R. Schrier. 1995. The global tobacco epidemic. Scientific American, May, 44–51. |
| 34. | Dodge R., Cline M. G., Burrows B.Comparisons of asthma, emphysema, and chronic bronchitis diagnoses in a general population sample. Am. Rev. Respir. Dis.1331986981986 |
| 35. | Vollmer W. M., Osborne M. L., Buist A. S.Uses and limitations of mortality and health care utilization statistics in asthma research. Am. J. Respir. Crit. Care Med.1491994S79S87 |
