Decrease in Incidence Rate of Hospitalizations Due to AIDS-Defining Conditions but Not to Non-AIDS Conditions in PLWHIV on cART in 2008–2018 in Italy

Background: We aimed to describe the change in the incidence and causes of hospitalization between 2008 and 2018 among persons living with HIV (PLWHIV) who started antiretroviral therapy (ART) from 2008 onwards in Italy. Methods: We included participants in the ICONA (Italian Cohort Naïve Antiretrovirals) cohort who started ART in 2008. All the hospitalizations occurring during the first 30 days from the start of ART were excluded. Hospitalizations were classified as due to: AIDS-defining conditions (ADC), non-ADC infections and non-infections/non-ADC (i.e., cardiovascular, pulmonary, renal-genitourinary, cancers, gastrointestinal-liver, psychiatric and other diseases). Comparisons of rates across time were assessed using Poisson regression. The Poisson multivariable model evaluated risk factors for hospitalizations, including both demographic and clinical characteristics. Results: A total of 9524 PLWHIV were included; 6.8% were drug users, 48.9% men-who-have sex with men (MSM), 39.6% heterosexual contacts; 80.8% were males, 42.3% smokers, 16.6% coinfected with HCV and 6.8% with HBV (HBsAg-positive). During 36,157 person-years of follow-up (PYFU), there were 1058 hospitalizations in 747 (7.8%) persons; they had HIV-RNA >50 copies mL in 34.9% and CD4 < 200/mmc in 27%. Causes of hospitalization were 23% ADC, 22% non-ADC infections, 55% non-infections/non-ADC (11% cancers; 9% gastrointestinal-liver; 6% cardiovascular; 5% renal-genitourinary; 5% psychiatric; 4% pulmonary; 15% other). Over the study period, the incidence rate (IR) decreased significantly (from 5.8 per 100 PYFU in 2008–2011 to 2.21 per 100 PYFU in 2016–2018). Age > 50 years, intravenous drug use (IDU), family history of cardiovascular disease, HIV-RNA > 50, CD4 < 200, were associated with a higher hospitalization risk. Conclusions: In our population of PLWHIV, the rate of hospitalization decreased over time.


Introduction
Combination antiretroviral therapy (cART) is effective in reducing morbidity and mortality due to HIV infection for people living with HIV (PLWHIV) [1]. Causes and 2 of 13 rates of hospitalization depend on considered settings. For example, people with HIV in New York City were more frequently hospitalized than people without HIV; while AIDS-defining illnesses were relatively rare, non-AIDS-defining infection hospitaliza-tions were more common [2]. In Sierra Leone, PLWHIV accounted for a substantial proportion of admissions in hospitals for AIDS-defining conditions, with a high mor-bidity and mortality burden, due to late HIV diagnosis [3].
In Italy in 2018, the proportion of PLWHIV with CD4+ cells count < 200/µL at diagnosis was 37.8% [4]. The initiation of antiretroviral therapy in HIV-positive adults provides net benefits in the development of AIDS-defining and non-AIDS-defining illnesses [5]. Recent data highlight the importance of considering and screening non-AIDSrelated conditions [6], which are the principal causes of admission, involv-ing all hospital departments [7].
The observation of causes of hospitalization in PLWHIV is important in our clini-cal practice. The use of antiretroviral therapy decreased AIDS-related events; on the other hand, we observed an increase in median age in PLWHIV.
Our hypothesis was that in the last 10 years, the use of cART with high immunovirological success was associated with a decrease in hospitalization rates and with a change of causes, with an increase in non-infectious and non-AIDS-related comorbidi-ties, in particular related to age and metabolic parameters.
This study describes incidence rate, causes of hospitalization, factors associated with the risk of hospitalization between 2008 and 2018 of people living with HIV starting ART after 2007.

Study Population
This is a retrospective analysis of prospectively collected data from the ICONA (Italian Cohort Naïve Antiretrovirals) Foundation cohort. ICONA is an observational cohort, set up in 1997, including HIV-1-infected subjects, naïve from ART at the time of enrollment. Demographic, viro-immunological and clinical data and information on antiretroviral regimens are collected and recorded using an electronic database.
All patients signed consent forms to participate in the ICONA Foundation Study in their local participating clinical sites. The research study protocol has been approved by local institutional review boards We included HIV-infected persons who started first antiretroviral therapy in the period from 1 January 2008 to 31 December 2018, having at least one follow-up visit. For those patients, a hospitalization was considered if it occurred at least 30 days after the enrollment and was defined as admission to any Italian hospital for two or more days. Participants' follow-up accrued from the date of enrollment (baseline) to their last visit or death or 31th December, 2018 (whichever was earlier). Participants could be included in multiple periods and could contribute more than one hospitalization per period. Socio-demographic factors were collected at baseline, such as sex, age group, nationality, HIV risk factor, family history of cardiovascular disease; HIV-related factors have been collected at baseline and during follow-up (i.e., CD4, HIV viral load, hepatitis C coinfection and time from HIV diagnosis to first ART).
The outcomes evaluated in this study were the rates of hospitalizations for any reason and hospitalizations for specific reasons grouped as follows: AIDS-defining conditions (ADC) (see Supplementary Table S1), infections non-AIDS-defining (including all the infectious parasitic diseases with the exclusion of those classified in ADC group, plus bacterial pneumonia, cellulitis and sepsis, see Supplementary Table S1), and noninfections/non-ADC (i.e., cardiovascular, pulmonary, renal/genitourinary, non-AIDS cancer, liver/gastrointestinal, psychiatric and other diseases, including all other causes not classified in the previously specified categories). We also assessed possible factors associated with hospitalizations.

Statistical Analysis
The baseline for this analysis was the date of the start of the first antiretroviral therapy. Study period hospitalization rates were calculated as the number of hospitaliza-tions occurring during the period divided by the total of person-years of follow-up (PY). For each participant, the PY at risk were computed from the starting day of the first ART to the last follow-up or death or 31 December, 2018 (whichever was earli-er). Participants could be included in multiple time periods and could contribute more than one hospitalization per time period.
The number of hospitalizations, PY at risk and incidence rate (IR) of hospitaliza-tion (per 100 PY) were calculated overall and for three study periods (every four years and the last period has three years): 2008-2011, 2012-2015 and 2016-2018. Crude and adjusted incidence rate ratios (IRR) and 95% confidence intervals were calculated using Poisson regression models, and the overdispersion of data was as-sessed with deviance and Pearson goodness-of-fit tests.
All factors included in this analysis were also evaluated using a more parsimoni-ous model-building strategy (i.e., inclusion of factors with a log-likelihood ratio test p-value < 0.2 at univariate analysis and then backward selection with a log-likelihood ratio test p-value < 0.2 at each step). However, we decided to report the model with all covariates, given that the results were similar when using this approach (data not shown).
Moreover, we performed the same analysis using hospitalization or death as the outcome.
Aiming to evaluate an effect modification of each factor in the different study pe-riods, we performed models with interaction terms between each factor and study pe-riods only for any reason of hospitalization.
All the analyses were performed with STATA (version 15).
At hospitalization 34.9% had HIV-RNA > 50 copies mL and 27% CD4 < 200/mm c . The highest number of hospitalization was due to non-infection/non-ADC (55%), 23% of hospitalizations were due to ADC and 22% to non-AIDS-defining infections.
The distribution of all the causes making up the definitions of ADC, infections non-AIDS-defining, and non-infections/non-ADC, over time are reported in Supplementary  Table S2.
The distribution of all the causes making up the definitions of ADC, infections non-AIDS-defining, and non-infections/non-ADC, over time are reported in Supplementary, Table S2.  (Table 2).   (Table 2).
Further when we performed models with interaction terms between each factor and study period, we found significant interactions only for CD4, viral load and mode of infection. However, the effects did not change the direction of the association but the magnitude was increasing as the study period was more recent.
When we performed the same analysis using hospitalization or death as the outcome, we obtained consistent results with those using as the outcome of hospitalizations (see Supplementary Table S3).
The hospitalizations due to ADC were associated with higher levels of viral load, particularly 51-10,000 and >10,000 copies/mL (IRR 1.97 and 2.75, respectively); the noninfection/non-ADC were most associated with other nationality than Italian (IRR 1.45), IDU and MSM (IRR 1.77 and 1.46, respectively) and high viral load, 51-10,000 and >10,000 copies/mL (IRR 1.69 and 1.77, respectively); be over 52 years old, IDU, have a family history of cardiovascular disease and viral load between 51 and 10,000 copies/mL appeared to be associate with a higher risk of hospitalization for infection non-AIDS-defining (IRR 2.02, 1.87, 1.66 and 1.47, respectively). Higher CD4 levels (≥200 cells/mmc) were protective for all reasons (Table 4).

Discussion
The current study showed a temporal decrease in the overall hospitalization rate and a decrease in the risk of hospitalization in people living with HIV in care from 2008 to 2018 in the ICONA cohort in care from 2008 to 2018 in the ICONA cohort. After the introduction of cART, the hospitalizations rate declined in PLWHIV, and this decline (−61.8%) was greater than the decline in age-standardized hospitalization rates observed for the general population in the same time period (−26.4%), according to the official register of Italian Health Departement. Non-AIDS-related hospitalizations were more frequent than those related to AIDS [2,[6][7][8][9].
A recent editorial underlines the importance of reaching zero hospitalizations to reach zero deaths attributable to HIV [10] and considers every hospitalization as a failure of public health.
When considering different reasons for hospitalization, major declines were detected for AIDS-defining events. This is comparable to other cohorts, and it is due to ART efficacy.
Hospitalizations due to non-infection/non-ADC were stable over time. Published data of the ICONA cohort showed an increase in the prevalence of some non-communicable diseases from 2004 to 2014; in particular, the prevalence of dyslipidemia, hypertension and cardiovascular disease increased [11]. In our study, cardiovascular disease caused 6% of hospitalizations. Oncologic disease caused 11% of hospitalizations; the malignancies in HIV Italian cohorts were observed in 8.5% of the CISAI cohort [12] and in 3.8% of the ICONA cohort [13].
The percentage of hospitalizations due to non-AIDS-defining infections increased during the observation time. The most common condition associated was pneumonia; this observation is consistent with another cohort [14] data; in our cohort, the crude incidence of pneumonia was 5.66 per 1000 person-years from 1996 to 2011 [15]. The second cause of hospitalization was a sexually transmitted infection; in Italy, there was an outbreak of sexually acquired hepatitis A [16], and there is a treatment failure to syphilis in about 10% of PLWHIV [17]; these data could explain the high rate of hospitalizations due to STDs.
The higher risk of hospitalization was associated with older age, use of injecting drugs and detectable viral load. The overall prevalence of multimorbidity in the geriatric Italian GEPPO cohort amounted to 64%, and it is related to longer duration of HIV infection rather than older age per se [18]; the prevalence of comorbidities in PLWHIV is higher than in HIV-negative people, probably due to the inflammation [19]. Early determination of frailty could lead to target interventions to decrease adverse outcomes that cause hospitalization.
The use of injecting drugs could be related to sepsis; the likelihood of bacteremia was found to have increased in a consortium of sites in the USA in the HAART era [20]. In our study, hospitalizations were found more frequently in the group of people with injecting drug use (former or current), not because of active injecting drug use.
Detectable viral load is associated with a higher risk of hospitalization and, in particular, to non-AIDS-defining events due to inflammation-related events [21].
Focus on social factors, low educational grades and unemployed are associated with a higher risk of hospitalization. Several studies have identified limited literacy as a risk factor for poor HIV medication adherence, despite evidence that questions this association [22,23]. These findings are in line with this issue.
This study has some limitations: first, ICONA is not representative of all Italian centers, and some of them are the reference for many clinical and surgical specialties; this could overestimate the hospitalization rate. Another limitation is the classification of hospitalization, with some diseases that could be misunderstood (for example, acute hepatitis infections could be classified as gastrointestinal diseases and sexually transmitted diseases). Moreover, it is of note that we estimated some IRR with very large 95% confidence intervals. This reflects the lack of statistical power for some of the potential determinants.
Our study is useful to understand the changing of hospitalization causes and rates in the era of antiretroviral therapy; we analyzed data of patients treated in the last 10 years considering the high rate of immunovirological success of recent cART, and in this setting, we conclude that prevention strategies against non-communicable diseases and non-AIDS-defining infections are crucial to decrease hospitalization rate. These strategies could include stop smoking, starting a low-calorie diet and use of specific drugs (e.g., statins, hypoglycemic).

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/ 10.3390/jcm10153391/s1, Table S1: ICD9 codes used to group hospitalizations, Table S2: Distribution of all the causes making up the definitions of ADC, infections non-AIDS defining, and non-infections/non-ADC over time, Table S3: Multivariable analysis of factors associated with hospitalization or death-Poisson regression model.