Rate of hospitalization (per 1000 US population) for all causes, and rate and proportion of hospitalizations for infectious diseases (per 1000 US population), 1980 through 1994. From National Hospital Discharge Survey data.
Hospitalizations for infectious diseases by age group, 1980 through 1994. Plus sign indicates upward trend; minus sign, downward trend. From National Hospital Discharge Survey data.
Point estimates of fatality among persons hospitalized with infectious diseases and other conditions, by age group, 1980 and 1994. From National Hospital Discharge Survey data.
Simonsen L, Conn LA, Pinner RW, Teutsch S. Trends in Infectious Disease Hospitalizations in the United States, 1980-1994. Arch Intern Med. 1998;158(17):1923-1928. doi:10.1001/archinte.158.17.1923
A recent study concluded that between 1980 and 1992, deaths from infectious diseases increased 58%. This article explores trends in infectious diseases as a cause of hospitalization.
We analyzed data from the National Hospitalization Discharge Survey for 1980 through 1994 using a previously developed approach to evaluate infectious diseases in data coded according to the International Classification of Diseases, Ninth Revision .
Between 1980 and 1994, the rate of hospitalizations in the United States declined approximately 33%; hospitalizations occurred at a rate of 133±5 per 1000 US population (35 million±1 million discharges) in 1994. The rate of hospitalization for infectious diseases declined less steeply—12% during this interval—resulting in an increased proportion of hospitalizations because of infectious diseases. In 1994, the rate of hospitalizations for infectious diseases was 15.4±0.7 per 1000 US population (4.0 million±0.2 million discharges). The fatality rate associated with hospitalizations for infectious diseases increased from 1.9%±0.1% to 4.0%±0.3%, attributable to increased hospitalizations of elderly persons and an increased fatality rate among those younger than 65 years. Among selected categories, hospitalizations for human immunodeficiency virus infections and acquired immunodeficiency syndrome, prosthetic device infections, sepsis, and mycosis increased substantially, and hospitalizations for upper respiratory tract infections, pelvic inflammatory disease, and oral infections declined sharply. Hospitalizations for lower respiratory tract infections constituted 37% of all infectious disease hospitalizations in 1994.
Considering hospitalizations as a dimension of the burden of infectious diseases involves an array of factors: secular trends in hospitalization, changing case management practices, demographic changes, and trends in the variety of infectious diseases themselves. Increases in the proportions of hospitalizations because of infectious diseases during years when hospitalizations for all causes were decreasing reflect an increasing burden of infectious diseases in the United States between 1989 and the mid-1990s.
IN 20TH-CENTURY United States and other postindustrialized societies, morbidity and mortality from infectious diseases have tended to decrease in contrast to morbidity and mortality from other conditions.1,2 However, numerous recent events have focused attention on the emergence and reemergence of infectious diseases.3- 7 Recently, we developed a method to evaluate the burden of infectious diseases in sources of data coded according to the International Classification of Diseases, Ninth Revision (ICD-9).8 We used this method to evaluate infectious disease mortality in the United States. From 1980 to 1992, infectious disease–related deaths increased 58% in the United States.9
We used a similar approach to study recent trends in hospitalizations for infectious diseases using data from the National Hospital Discharge Survey (NHDS), which includes a representative sample of hospital discharges recorded in the United States. An important challenge for this analysis has been understanding trends in hospitalizations for infectious diseases in the context of recent changes in the delivery of health care, including an overall trend toward fewer hospitalizations.
We analyzed public-use tapes of data for 1980 through 1994 from the NHDS, which is maintained by the Centers for Disease Control and Prevention's National Center for Health Statistics, Hyattsville, Md.10,11 These tapes contain approximately 200000 discharge records per year from a survey sample of short-stay, nonfederal hospitals located in the United States. The survey is based on a stratified, multistage probability design; the sampled hospital discharge records are weighted to produce national estimates. The database includes demographic information, up to 7 discharge diagnoses per hospitalization, number of days of care, and vital status at discharge. With a few exceptions, NHDS tapes tabulate discharge diagnoses in the order they are listed on the abstract form.10- 12
When not otherwise specified, a hospitalization for infectious disease was defined as a discharge record with an infectious disease ICD-9 code as the first-listed diagnosis. The first-listed diagnosis is the one specified as the principal diagnosis on the face sheet or discharge summary of the medical record or the diagnosis listed first if the principal diagnosis is not specified.10- 12
In addition to the first-listed diagnosis, we also used as a definition of hospitalization for infectious disease a discharge record with any mention of an infectious disease ICD-9 code among the up to 7 listed discharge diagnoses (multiple cause).
We included all discharges in the database, including routine hospital stays for newborns.
We used a protocol previously developed for infectious disease mortality to classify ICD-9–coded diagnoses according to whether they are infectious diseases, and we restricted this analysis to those ICD-9 codes that indicated an infectious disease in almost all cases.9
Using statistical analysis software (SAS, SAS Institute, Cary, NC), we computed the estimated number of hospitalization discharges (1) assigned to an infectious disease ICD-9 code as the first-listed diagnosis or (2) with any mention of an infectious disease among the up to 7 discharge diagnoses (ie, multiple cause). Unless specifically described as multiple-cause rates, the reported numbers and rates are based on the first-listed diagnosis.
We examined hospitalizations for 6 age groups (0-4, 5-24, 25-44, 45-64, 65-84, and 85 years and older) and for males vs females. Data on race were dichotomized (whites vs blacks and all other races). We used the variable of vital status at discharge to compute the fatality rate associated with hospitalizations.
We arranged the infectious disease ICD-9 codes into 21 groups of related conditions, which accounted for about 93% of discharges associated with infectious disease in 1980 and 1994 (the remaining 7% of discharges associated with infectious disease diagnoses were not specific enough to be included in one of these disease groups) (Table 1). Each disease group was evaluated for trends in hospitalizations.
For each year, we estimated the number of hospital discharges for infectious disease, the number of days of care, and the fatality rate. We present SEs that were estimated using variables and equations for the relative SE produced by the National Center for Health Statistics for use with the NHDS, which used a computerized routine based on a rigorous, unbiased algebraic estimator of the variance for 1980 to 1987 data and statistical analysis software (SESUDAAN, Research Triangle Institute, Research Triangle Park, NC) for 1988 through 1994 data.10- 12 The approximate SEs in percentage estimates, in which the numerator either was or was not a subclass of the denominator, were computed with appropriate formulas derived by the National Center for Health Statistics for this purpose.13
We used weighted linear regression to assess trends for each of 15 point estimates (1980-1994) in the number and rate of hospitalizations and hospitalizations for infectious diseases and in the number of hospitalizations for each of the selected infectious disease groups. Weighting was done by the inverses of the variances of the point estimates. Estimates of the average annual change in the number of hospitalizations were based on the slope of the regression line; relative changes among the various infectious disease groups were characterized by the average annual change in number of hospitalizations divided by the number of 1987 hospitalizations (the midpoint year for the period studied). The determination of statistical inference is based on t13=2.16 (P=.05).13
Between 1980 and 1994, the rate of hospitalizations in the United States declined approximately 33%; in 1994, hospitalizations occurred at a rate of 133±5 per 1000 US population (35 million±1 million discharges). The rate of hospitalization for infectious diseases declined less steeply: 12% during this interval (Figure 1). In 1994, the rate of infectious disease hospitalizations was 15.4±0.7 per 1000 US population (4.0 million±0.2 million discharges). As a consequence of these differing rates of change, the proportion of hospitalizations because of infectious diseases increased approximately 27%; in 1994, infectious diseases accounted for 12%±0.2% of hospitalizations.
Between 1980 and 1994, the number of hospital days of care for all causes declined significantly; in contrast, the number of days of hospitalization for infectious diseases increased significantly. Thus, the proportion of hospitalization days attributable to infectious disease as the first-listed cause was 8.8%±0.3% in 1980 and 13.9%±0.1% in 1994. The rate of hospitalization days for all causes declined during the study interval (1275±45 days per 1000 US population in 1980 and 717±30 days per 1000 population in 1994). However, the rate of hospitalization days for infectious diseases did not change significantly (113±6 days per 1000 US population in 1980 and 100±4 days per 1000 population in 1994).
Trends in hospitalizations for infectious disease from 1980 through 1994 differed among age groups (Figure 2). Rates of hospitalization in 65- to 84-year-olds and those 85 years and older increased significantly, whereas rates in persons aged 25 to 64 years and those younger than 25 years decreased significantly.
Overall, hospitalization rates for infectious disease did not differ by sex in 1994 (14.9±1.0 per 1000 males and 15.8±0.6 per 1000 females). However, between 1980 and 1994, trends in infectious disease hospitalizations differed between males and females by age group. Rates declined in those younger than 25 years in both sexes (39% in males and 53% in females) and increased among those older than 65 years in both sexes (44% in males and 50% in females). However, in 25- to 64-year-olds, infectious disease hospitalization rates declined significantly in females (31%) but did not change in males.
The proportion of all hospitalizations that were for infectious diseases was 11% for whites and 12% for blacks and all other races in 1994. Approximately 20% of all hospitalization records in the 1994 NHDS data set did not include information about race, and no further analysis of race-specific data was done.
The fatality rate for patients hospitalized for infectious disease as the first-listed discharge diagnosis increased from 1.9%±0.1% in 1980 to 4.0%±0.3% in 1994 (Figure 3). In contrast, the fatality rate for other (noninfectious) hospitalizations remained unchanged at approximately 2.4%±0.2%. For persons 65 years and older, the fatality rate among those hospitalized for an infectious disease was 7.2%±0.5% in 1980 and 7.4%±0.3% in 1994; for persons younger than 65 years, the rate was 0.5%±0.1% in 1980 and 1.9%±0.2% in 1994. Among persons younger than 65 years, human immunodeficiency virus infection and acquired immunodeficiency syndrome (AIDS) accounted for approximately 2% of infectious disease hospitalizations (although it appeared as one of the diagnoses 5 times as often as it was listed as the first diagnosis); in this age group, human immunodeficiency virus infection and AIDS had a 10% fatality rate. For hospital discharges related to noninfectious disease, the fatality rate was significantly lower in 1994 than in 1980 among persons 65 years and older (6.9%±0.3% in 1980 and 5.1%±0.1% in 1994) and was comparable for persons younger than 65 years (1.0%±0.1% in 1980 and 0.8%±0.2% in 1994).
In contrast to decreases in the number of hospitalizations, the total number of discharge diagnoses listed for all hospitalizations was an estimated 28% higher in 1994 (125 million total diagnoses) than in 1980 (98 million total diagnoses). Consequently, the average number of listed discharge diagnoses per hospitalization was also higher: 3.6 in 1994 vs 2.3 in 1980. For hospitalizations that included an infectious disease diagnosis, an average of 4.7 diagnoses were listed in 1994 compared with 3.0 in 1980. The proportion of hospitalizations in which an infectious disease appeared among the diagnoses was 16% (6.5 million of 42 million) in 1980 and 22% (7.6 million of 35 million) in 1994.
Trends in hospitalizations varied substantially by infectious disease group (Table 2). For example, hospitalizations for septicemia, infections caused by prosthetic devices, AIDS, and postoperative infections increased more than 9%. In contrast, hospitalizations for upper respiratory tract infections, pelvic inflammatory disease, and oral cavity infections decreased more than 8%. Hospitalizations for lower respiratory tract infections constituted 37% of all hospitalizations for infectious disease in 1994.
From 1980 through 1994, the rate of hospitalizations declined by approximately one third in the United States, and the rate of hospitalizations for infectious diseases declined only 12%. Consequently, the proportion of all hospitalizations that were because of infectious diseases increased approximately 27%, reaching 12% of all hospitalizations in 1994. If there were less discretion in decisions about hospitalizing persons with infectious diseases than with other conditions, this increasing proportion of infectious disease hospitalizations could be solely a consequence of general trends toward fewer hospitalizations. Another possible explanation is that the burden of infectious diseases has been increasing in recent years in a climate favoring fewer hospitalizations, and, as a result, infectious diseases hospitalization rates declined somewhat, but an increasing proportion of hospitalizations was because of infectious diseases.
Some data presented here suggest that the latter possibility, an increasing burden of infectious diseases, contributed to these trends. First, the fatality rate among persons hospitalized with infectious diseases more than doubled from 1980 to 1994. In contrast, among all persons hospitalized with noninfectious conditions, the fatality rate did not change appreciably. Approximately two thirds of the increase in the number of deaths among persons hospitalized with an infectious disease occurred in those older than 65 years, in whom there was no change in fatality rate but an 83% increase in the number of infectious disease hospitalizations. In contrast, one third of the increase occurred in those younger than 65 years, despite a 22% decrease in infectious disease hospitalizations, because the fatality rate nearly quadrupled (in part because of human immunodeficiency virus infection and AIDS). Second, although the number of days of hospitalization for all causes decreased, the number of days of hospitalization for infectious diseases increased. These increases in fatality rate and number of hospitalization days suggest that hospitalization for persons with infectious diseases is increasingly reserved for the sickest among those with infectious diseases and that if practices had remained constant, the rate of hospitalization for infectious diseases would have increased substantially.
Changes in practices of discharge diagnosis coding might have affected the patterns observed. Diagnosis-related groups, introduced in 1983, and the associated changes in reimbursement policies created an incentive for choosing discharge diagnoses with higher reimbursement rates, which could have had an impact; we did not evaluate this possibility directly.
Analysis of the data on multiple causes of hospitalization shows an increasing number of diagnoses between 1980 and 1994, even as the number of hospitalizations was decreasing. This trend could reflect hospitalization of sicker persons or simply a tendency toward listing more diagnoses. Hospitalizations with an infectious disease as one of the diagnoses have a higher number of diagnoses than other hospitalizations; in addition, the average number of diagnoses for these hospitalizations was 57% higher in 1994 than in 1980 (4.7 vs 3.0). Infectious disease is often a complication of other conditions (ie, an additional diagnosis), a factor that may explain this observation. Because of these complications in evaluating the data on multiple causes of hospitalization, and to provide a conservative estimate of the impact of infectious diseases on hospitalizations, we focused on the first-listed diagnosis. In addition, this analysis focused only on those diagnosis codes that represent an infectious disease in almost all cases.
Trends in hospitalization for specific infectious disease diagnoses varied considerably, reflecting an assortment of developments. Increasing hospitalizations for AIDS reflects the increasing incidence of that disease, and the increases for prosthetic device infections may reflect the increasing use of these devices (not necessarily an increased per device infection rate). In contrast, hospitalizations decreased substantially for upper respiratory tract infections, pelvic inflammatory disease, and oral infections, perhaps reflecting a tendency toward outpatient management of these conditions.
For certain diagnoses, estimates of the burden of disease differed depending on whether only the first-listed or all diagnoses were counted. For example, the diagnoses of AIDS and mycoses were listed among all diagnoses more frequently than they were the first-listed diagnosis (5 and 10 times, respectively); using the first-listed diagnosis alone underestimates the burden of these infections.
Considering hospitalizations as a dimension of the burden of infectious diseases involves a complicated array of factors in addition to trends in infectious diseases themselves. A general move toward fewer hospitalizations has had an impact on infectious disease hospitalizations, although the proportion of hospitalizations because of infectious diseases has increased. Demographic changes, including an aging population and longer survival of persons with a variety of medical conditions who are more susceptible to infectious diseases, have affected trends in infectious disease hospitalizations. Current medical technologies, which have contributed to improved survival, may also have increased the opportunity for infections through increased use of indwelling intravascular and other invasive devices. This analysis of trends in hospitalizations for infectious diseases extends a previous analysis of infectious disease deaths.3 The picture emerges from both studies of an increasing burden of infectious diseases in the United States between 1980 and the mid-1990s.
Accepted for publication January 22, 1998.
We thank Robert Pokras, Lola Jean Kozak, PhD, and Maria Owing, PhD, Hospital Care Statistics Branch, National Center for Health Statistics, Centers for Disease Control and Prevention, for carefully reviewing earlier drafts of the manuscript and for assisting in interpreting survey sample estimates; Iris Shimizu, PhD, Office of Research and Methodology, National Center for Health Statistics, for accessing and generating NHDS data sets; Meade Morgan, PhD, and Bao-Ping Zhu, MD, MS, for providing advice on statistical analysis; and Tara Strine, MS, and Matthew J. Clarke, MS, Biometrics Branch, Centers for Disease Control and Prevention, for providing programming advice.
Reprints: Robert W. Pinner, MD, OD/NCID/CDC, Mail Stop (C12), 1600 Clifton Rd NE, Atlanta, GA 30333.