Trends in hospital admissions at a Department for Infectious Diseases in Italy from 1995 to 2011 and implications for health policies

Background Interactions among several environmental, behavioral, social, and biological variables contribute to the epidemiology of infectious diseases (IDs) and have an impact on the healthcare system and hospitalizations. We evaluated trends in ID hospitalizations at our Department for Infectious Diseases in the last two decades to aid decision-makers in defining appropriate healthcare strategies. Methods The discharge diagnoses of all patients admitted to the ID Department of L Sacco University Hospital between 1995 and 2011 were classified by the International Classification of Diseases (ICD-9) grouped in Major Diagnostic Categories (MDC). Linear regression was used to determine the trends in hospitalizations for each MDC. Estimates of the average annual change were based on the slope of the regression line. Results A sharp decline in HIV/AIDS cases (−22.5 +/−6.0 cases per calendar year), and an increase in admissions for respiratory, cardiovascular, renal and musculoskeletal infections were recorded. The mean age of the patients increased by 1.2 years (+/−0.049) for each calendar year of observation (linear trend, p < 0.0001), increasing from 37.02 +/−11.91 years in 1995 to 56.02 +/−19.62 years in 2011 (p < 0.0001). The mean number of comorbidities per patient increased significantly over time (Mann–Whitney U test, p = 0.0153). From 1998/1999 to 2010/2011 the hospital length of stay (LOS) increased for cardiovascular, digestive system, musculoskeletal, and skin/subcutaneous infections, and infectious and parasitic diseases (p < 0.01). The rate of hospital stay over threshold (HSOT) increased in the last 5 years by 1.12% for every 10-year age group. Conclusions Older age, a higher number of comorbidities, a longer hospital LOS for certain conditions, and a higher rate of HSOT characterize the patients admitted to this ID department in recent years. Despite progress in treatment and management, infectious diseases continue to be a major threat to human health. The current challenge for ID departments is the treatment of complex cases, often associated with chronic diseases in elderly patients. Continuous monitoring at a local and national level will allow early identification of changes in the epidemiological patterns of IDs and provide information for healthcare system planning.


Background
The epidemiology of infectious diseases (IDs) is a composite of several environmental, behavioral, social, and biological variables, including the selective effect of drugs or vaccines, and other unpredictable events that can change over time and have potentially important effects on global health. It is estimated that IDs are the second leading cause of death worldwide after cardiovascular diseases [1]. Respiratory infections, diarrheal diseases, human immunodeficiency virus acquired immunodeficiency syndrome (HIV/AIDS), tuberculosis (TB), and malaria account for 18.3% of all causes of death in 2004, although major differences in their ranking exist between high-and low-income countries [1].
Communicable diseases remain a major health threat in Europe, although there are wide differences among the World Health Organization European regions. In the "Health at a Glance: Europe 2012" more than 6,000 new hepatitis B cases, with a mean rate of 2.0/100,000 population, and 27,000 newly diagnosed HIV infections were reported by EU member states [2].
The public health importance of IDs and communicable diseases is mostly related to the continuing threat of epidemic/pandemic events), the emergence of new diseases and re-emergence of old diseases, the threat of imported diseases or pathogens, and the emergence of multidrug or pan-drug resistant organisms [3][4][5][6][7][8][9][10][11][12][13].
Hospital ID departments are a useful source of surveillance data that can indicate changing healthcare requirements. For example, a great effort was made In Italy in the late 1980s to tackle the AIDS epidemic, with an increase in the overall number of hospital beds to deliver services to HIV-infected inpatients. Twenty years on, the healthcare needs of HIV-infected patients have changed, and they are now mainly followed up as outpatients. In 2010, the overall rate of hospital discharge of HIV/AIDS patients was 3.9/10,000 compared with 29.87/10,000 inhabitants for other IDs excluding AIDS (ICD-9 codes 001-139) [14]. Over the same period, however, there has been an increasing need for hospital admission for several other infectious or communicable diseases.
The rapid and often unpredictable changes in the epidemiology of IDs, the significant burden on public health and the global economy, the breakthroughs in the field of prevention, and the new treatment opportunities for acute and chronic infectious diseases require ongoing evaluation in order to formulate appropriate healthcare and public health strategies. In particular, several structural and organizational choices for ID departments and related services must be made in relation to changing healthcare needs and priorities.
This study determined the temporal trends of hospitalizations in an ID department in the period 1995-2011 after the introduction of hospital funding on a "per case" basis (1995), classified according to the Diagnosis Related Group (DRG) system in Italy. The aim of the study was to evaluate changes in healthcare needs of patients admitted to an ID department and to provide some evidence for structural and organization changes in healthcare delivery for patients suffering from infectious or communicable diseases.

Methods
We retrospectively analyzed the discharge diagnosis of all patients admitted to the ID Department of L Sacco University Hospital between January 1995 and December 2011. This Department is a referral center for acute and highly communicable infectious diseases in Northern Italy. It provides the clinical care and follow-up of patients older than 18 years suffering from viral, bacterial, and parasitic diseases. Patients affected by gynecological infections are admitted to Gynecology/Obstetrics Departments. The ID department includes outpatient wards and day hospital services, a specific emergency room, and wards for hospital admissions (87 beds). The ID specialists on 24-hour duty at the Emergency Department of the Hospital decide on the admission of patients with suspected infectious or communicable diseases to their own Department according to triage rules, and to epidemiological and/or clinical priorities. Patients who need intensive care are primarily admitted to the intensive care unit and not to the ID department.
From the patient records, we collected demographic information, up to six discharge diagnoses and up to six procedures per hospitalization, the DRG, and the number of days of in-hospital care. The ICD-9 codes were classified into 25 groups (Major Diagnostic Categories (MDC) version 24) of related conditions which accounted for all the DRGs at discharge, and the MDCs were evaluated for their trends over time.
As surrogate markers of clinical complexity we used the number of ICD-9 codes for diseases (ICD-9c) and procedures (ICD-9p) notified in the hospital discharge card, the hospital length of stay (LOS), and the hospital stay over threshold (HSOT; days of hospitalization over the limit for the specified condition defined by national average data).
We examined temporal trends of hospitalization according to the demographic characteristics of the population: sex, age, and age group.

Statistical analysis
The D' Agostino and Pearson omnibus normality test was used for the descriptive variables to assess their distribution. The unpaired t-test with Welch correction and the Mann-Whitney U test were used to compare the mean or medians of parametric and non-parametric values respectively. A p-value < 0.05 was considered statistically significant.
Linear regression was used to estimate the trends in hospitalization for each MDC according to sex, age, geographical origin of the patients, number of ICD-9c and ICD-9p, and in-hospital LOS. Estimates of the mean annual change were based on the slope of the regression line. Analyses were conducted with the GraphPad Prism version 4.02 for Windows (GraphPad Software, San Diego, CA, USA).
Multivariate linear regression analysis was performed for hospital LOS, including age group and sex in the model. The association between HSOT and demographic and clinical characteristics was evaluated by logistic regression. To account for the effects of several potential confounding factors, we used unconditional multiple logistic regression, with maximum likelihood fitting, to obtain odds ratios and their corresponding 95% confidence intervals. Included in the regression equations were sex, geographical origin of the patients, age group, and number of ICD-9c and ICD-9p. Multivariate analyses were performed using the SAS/STAT software package version 9.1 (Cary, NC, USA).
The study was approved by the local Ethical Committee on July 2013 (Protocol N 509/2013/73/ap) and was conducted in accordance with the Helsinki declaration.

Results
Between January 1995 and December 2011, there were 26,253 admissions to the ID department. The mean number of annual hospital admissions was 1544 ± 200.4. In total, there were 17,719 (67.49%) males and 8,534 females; the median age was 42.4 years (interquartile range (IQR): 34.3-57.7 years); median hospital LOS was 10 days (IQR: 6-19 days). Foreign-born patients accounted for 19.67% of the total admissions. Table 1 presents the crude number of admissions for specified conditions in the whole period, in the first and last year of observation, the slope of linear regression for trend, and its deviation from zero. Overall, the number of admissions for MDC-25 (HIV/AIDS) was by far the highest (44.4%), followed by MDC-07 (diseases and disorders of liver, gallbladder and pancreas) (12.68%), MDC-18 (infectious and parasitic diseases) (11.34%), and MDC-04 (diseases and disorders of the respiratory system) (9.96%). Within MDC-07, acute and chronic viral liver diseases (DRG 205, 202, 206) accounted for 87.14% of the cases; within MDC-18, infectious diseases, sepsis, viral infections and fever of unknown origin (DRG 423, 416, 421, 420, 576) accounted for 89.89% of cases; within MDC-04, pneumonia, other respiratory infections, and chronic obstructive pulmonary diseases (DRG 89, 90, 79, 80, 88) accounted for 82.52% of the cases.

Trend in hospital admissions by MDC
In the model adjusted for sex and age (>45 years) a significant increase in admissions for infectious diseases of the ear, nose, mouth and throat (MDC-03), respiratory infections (MDC-04), infections of the cardiovascular system (MDC-05), infections of musculoskeletal/connective tissue (MDC-08), infections of the kidneys and urinary tract system (MDC-11), and unspecified infectious or parasitic diseases (MDC-18) was found over the period, while there was an overall decrease in HIV/AIDS admissions (−22.52 ± 6.0 cases per calendar year). All significant variations in hospital admissions were associated with age > 45 years except for MDC-18 (infectious and parasitic diseases), which had variations associated with female sex.
The graph of hospital admissions for HIV/AIDS showed a rapid decline after 1996 ( Figure 1), coinciding with the widespread introduction of highly active antiretroviral therapy (HAART). Nowadays, although HIV/AIDS remains the main cause of admission to the ID department, the proportion of admissions has fallen from 83.13% to 28.34% of all causes of admission, while the proportion of admissions for other MDC has progressively increased.

Demographic data
In 1995, when HIV/AIDS infection was the major cause of hospital admission to our ID department, the male/female ratio of admitted patients was 2.9:1. Over time a progressive narrowing of the gap was observed, for an increase in admissions of female patients (admissions slope ± standard deviation (SD), +19.05 ± 3.54; deviation from zero, p < 0.0001) while the admissions for males remained constant.
The number of foreign-born patients was stable during the period (slope, +7.82 ± 3.83; no significant deviation from zero) while the number of Italian-born patients increased by +13.28 ± 5.41 for each calendar year (deviation from zero, p = 0.027).
The mean age of the patients increased by +1.2 ± 0.049 years per calendar year (linear trend, p < 0.0001) from 37.02 ± 11.91 years in 1995 to 56.02 ± 19.62 years in 2011 (unpaired t-test with Welch's correction, p < 0.0001).
In the analysis by age group, the overall increase in the mean age was a result of two opposite trends: an increase in hospital admissions of the oldest patients (slope, +20 ± 1.7 patients/year for those aged >75 years and +12 ± 1.4 patients/year for those aged 65-74 years), and a decrease in admission of patients aged 25-34 years (slope, −30 ± 1.5 patients/year). In the last year of observation, the 45-54-year and the >75-year age groups were the ages most frequently admitted ( Figure 2).  high number of HSOT was observed for some common conditions: 81 cases (12.88%) for diseases of the musculoskeletal system and connective tissue (MDC-08), 247 cases (8.29%) for infectious and parasitic diseases (MDC-18) and 1996 cases (17.12%) for HIV/AIDS (MDC-25). For these conditions, the trend of HSOT tended to decrease for HIV/ AIDS and to increase for MDC-18 and for MDC-08 ( Figure 3a). The rate of HSOT was significantly related to age: in the last 5 years (2007-2011), the rate of HSOT increased by 1.12% for each 10-year age group and in the last year of observation this increase was particularly high (2.69% for each 10-year age group) (Figure 3b). After excluding HIV-infected patients, the proportion of patients with HSOT was 5.96%. The commonest primary diagnosis of patients with HSOT was liver cirrhosis, followed by other infectious and parasitic diseases and sepsis. Figure 4 illustrates the proportion of HSOT in patients with the 20 most common DRGs.    Patients with HSOT (excluding HIV/AIDS) were significantly older than patients who had a shorter hospital stay (median age: 59.12 years (IQR 39.8-74.23) vs 50.33 years (IQR 34.31-69.92). These patients also had a significantly higher number of ICD-9c and ICD-9p. In the univariate model, the odds ratio of HSOT was significantly higher for foreign-born patients, for patients older than 45 years, and for those with more than one ICD-9c or at least one ICD-9p on the discharge card. In the multiple logistic regression analysis, the number of ICD-9c and ICD-9p remained significantly associated with a longer hospital stay, while the geographical origin of the patient and the age group showed no significant association (Table 3).

Discussion
In the whole period considered a significantly higher rate of hospital admissions in the ID Department for MDC-25 (HIV/AIDS), MDC-07 (Diseases and Disorders of liver, gallbladder and pancreas), MDC-18 (Infectious and parasitic diseases), and MDC-04 (Diseases and disorders of the respiratory system) can be observed if compared with the rate of admissions registered in Italy in 2010 in all public hospitals so as they can represent specific conditions for the ID Departments [14].
This study showed a marked change in the type of ID which resulted in hospital admission during the period 1995-2011. The most striking change was the decrease in hospital admissions for HIV/AIDS from 84.4% of all hospital ID admissions in 1995 to 28.3% in 2011. This was primarily due to the changes in the clinical characteristics of HIV/AIDS after the introduction of HAART in 1996. The disease was initially characterized by acute and life-threatening opportunistic infections and cancers needing hospitalization in acute care hospitals; after the widespread use of HAART in the industrialized world, HIV/AIDS changed to a chronic disease mainly managed in outpatient settings. The HIV Research Network described a decrease in the all-cause hospitalization rate for persons living with HIV/AIDS from 35/100 personyears to 27/100 person-years from 2002 to 2007, and an increase in hospital admission rates for cardiovascular, renal and pulmonary conditions associated with HIV infection [15][16][17][18]. However, the drop in admissions for HIV infection was quickly replaced with an increase in admissions for other infectious diseases.
The second most important change observed in this study was the aging of the admitted patients. In the 1990s, the great majority of patients were young HIVinfected patients, but now patients are more likely to be middle-aged or old patients with chronic-degenerative comorbid conditions. The most marked increase in hospital admission was recorded for patients aged > 75 years, with a mean increase in ID admissions of more than 20 cases per year.
Aged patients and comorbidities could, in part, explain the increase over time of the complexity of admitted patients, with increases in the mean number of ICD-9 codes and procedures, hospital LOS, and rate of HSOT. One of the limits in the definition of the overall complexity of the patients admitted is the lack, in the hospital discharge card, of a reliable method for the evaluation of this parameter. It would be of interest to include the age-adjusted Charlson Comorbidity Index in the hospital discharge card, not only for clinical and epidemiological reasons, but also to adjust for the overall clinical complexity of cases the financial reimbursement system (DRG system). The decrease in LOS over the whole period is apparently in contrast with the higher clinical complexity measured by the number of ICD-9c and ICD-9p for each patient. The greatest difference was observed in the years 1996-1997 when the mean LOS fell from 27.13 days to 16.33 days and thereafter remained stable. The dramatic drop was related to the changes in healthcare funding for hospitalizations rather than to changes in clinical/epidemiological parameters. The Italian Finance Act No 724/ 94 introduced hospital funding on a "per case" basis from calendar year 1995. The mean adjustment period of the system to reach equilibrium was 3 years. After this period a significant decrease in the total number of in-hospital beds for acute diseases (−24.8% from 1995 to 2003), an increase in the activities volume, and a decrease in LOS was observed in all Italian public hospitals [19]. After 1997, we found that the overall LOS remained stable, but there were significant differences among MDCs: the hospital LOS increased significantly for MDC-05, −06, −08, −09, and −18, and decreased for MDC-07 and −25.
In the adjusted model for sex and age group, an increasing trend for infections of the cardiovascular system, and for infections and parasitic diseases, and a