Abstract
HIV positive patients are a high risk population due to the alteration in their immune status. Healthcare associated infections (HAI) have not been well described in this population, with some risk factors reported inconsistently in the literature. The aim of this study was to describe the epidemiology as well as the underlying risk factors for HAI, specifically urinary tract infection (UTI), bloodstream infection (BSI) and respiratory tract infection (RTI). This was a retrospective cohort study conducted in three hospitals at an academic health system in New York City, over a two year period from 2006 to 2008. There were 3,877 HIV positive patient discharges in 1,911 patients. There were a total of 142 UTI, 106 BSI, and 100 RTI. The incidence rates were 4.35 for UTI, 3.16 for BSI and 2.98 for RTI. CD4 count and antiretroviral therapy were not associated with HAI. Significant predictors of UTI included urinary catheter, length of stay, female gender, steroids and trimethoprim-sulphamethoxazole (TMP-SMX); of BSI were steroids and TMP-SMX; and RTI were mechanical ventilation, steroids and TMP-SMX. Multivariable analysis indicated that TMP-SMX was significantly associated with an increased risk of infection for all three types of HAI [BSI odds ratio 2.55, 95% confidence interval (1.22–5.34); UTI odds ratio 3.1, 95% confidence interval (1.41–7.22); RTI odds ratio 5.15, 95% confidence interval (1.70–15.62)]. HIV positive patients are at significant risk for developing HAI, but the risk factors differ depending on the specific type of infection. The fact that TMP-SMX is a risk factor in these patients warrants further research as this may have significant health policy implications.
Keywords: Health care associated infections, HIV, trimethoprim sulphamethoxazole
Background and significance
Healthcare associated infections (HAI) in human immunodeficiency virus (HIV) negative patients have been well described; however the literature about HAI in HIV positive patients is limited and discrepancies remain regarding risk factors for the development of HAI in this population. The incidence of HAI is approximately 5–11% in industrialised nations, (Pittet et al, 2008) with an estimated 1.7 million HAI occurring annually and approximately 99,000 deaths in the United States of America (USA) (Klevens et al, 2007). In 2008 an estimated 1.1 million individuals were infected with HIV in the USA according to the Centers for Disease Control and Prevention (CDC) (CDC, 2011). The burden of HAI is great in the immunocompetent host, but it may be even greater in the HIV subset population. Hence the risk of HAI in this population warrants further research.
HIV positive patients tend to be admitted more frequently to hospital due to their immunosuppressed state and this increases their risk of HAI (Padoveze et al, 2002). The resulting HAI can lead to longer hospitalisation, economic burden and higher risk of mortality (Tumbarello et al, 1998). The literature is disparate regarding certain risk factors for HAI in HIV individuals, especially the role of CD4 count in predicting infection. There are also discrepancies in the literature regarding causative organisms as well as the site of infection.
The aims of this study were to describe the epidemiology of HAI among HIV infected patients to determine incidence rates, risk of death from HAI, frequency of organ-specific sites involved and the microbiologic aetiologies of HAI; and to ascertain the specific risk factors for HAI in HIV infected patients, specifically evaluating the role of CD4 count in the development of HAI.
Methods
Sample and setting
The data for this study were drawn from an electronic database that was compiled for the purpose of determining the economic distribution of the costs of drug resistant infections at an academic hospital system located in upper Manhattan (Apte et al, 2011). The academic institution is an urban hospital system in New York City comprised of four hospitals – a paediatric hospital, a community hospital and two tertiary/quaternary hospitals which provide care to a diverse patient population. The data were collected between 2006 and 2008 and approximately 320,000 discharges were documented during this time.
The sample in this study is a subset of the original study population: all adult patients who were HIV positive. Patients less than 18 years old were excluded, as were HIV positive women who were admitted for childbirth. This resulted in approximately 3,877 discharges, which was the unit of analysis.
Data collection
Following approval from the Columbia University Medical Center Institutional Review Board, data extracted from the database included demographics; laboratory reports, including microbiological results from urine, blood and respiratory specimens; device utilisation, specifically urinary and central venous catheterisation as well as mechanical ventilation; patient medical history including medicinal and recreational drug use; inpatient records including total length of stay, admission to intensive care unit (ICU) and death. The identification of HIV status was based on the International Classification of Diseases-9-CM.
Study definitions
The HAI examined included urinary tract infection (UTI), bloodstream infection (BSI), respiratory tract infection (RTI) or a combination of these infections, which developed at least 48 hours after hospital admission. If a patient had been discharged and then re-admitted with an infection within 72 hours, this was also considered as a HAI. The definitions of a UTI, BSI and RTI were based upon those used in the ‘parent’ study (Apte et al, 2011). Briefly, a team of clinicians and researchers developed electronic algorithms to identify HAI of the specific organ systems. Surveillance definitions from the CDC National Healthcare Safety Network for HAI were used to identify factors of infections that could be mapped to electronic data (www.cdc.gov/nhsn/about.html). Patients were categorised as infected, non-infected and uncertain. Patients in the uncertain category for HAI were not included in the analyses in order to minimise misclassification bias.
The organisms examined in this study were six common organisms causing antibiotic resistant infections in the study hospitals and therefore analysed in the ‘parent’ study: Acinetobacter baumanii, Enterococcus faecalis/faecium, Klebsiella pneumoniae, Pseudomonas aeruginosa, Staphylococcus aureus, Streptococcus pneumonia (Apte et al, 2011).
Statistical analysis
Initially separate univariable analyses for UTI, BSI and RTI were conducted using generalised estimating equations (GEE) for independent predictors, discussed below, to assess if they were significant by estimating odds ratios, 95% confidence intervals and p values. A p value of <0.05 was determined to be significant in this analysis. All variables which were found to be significant at the p<0.05 level were analysed in a multivariable model.
The demographics variables examined as potential predictors of HAI were age and gender, laboratory data included admission CD4 counts (< 200 cells/ul or ≥ 200 cells/ul) whenever available. Device-related variables assessed as independent predictors were central venous (CV) line, urinary catheterisation, and mechanical ventilation. Information regarding length of stay prior to infection (LOS), total length of stay, corticosteroids (current or ever used) and antiretroviral agents (ARV) prior to development of HAI were also analysed as independent predictors.
Since trimethoprim-sulphamethoxazole (TMP-SMX) is used as long-term prophylaxis in HIV infected individuals, it was also included as a potential predictor of risk of HAI.
Comorbid conditions previously identified in the literature and independently examined as potential confounders included diabetes mellitus, chronic renal failure, chronic obstructive pulmonary disease (COPD), systemic lupus erythematosus, rheumatoid arthritis and Hodgkin’s disease; however the numbers for the latter three were very small and did not contribute to the overall analysis.
The inpatient mortality was also assessed to determine if those patients who developed a HAI were at an increased risk of death compared with the HIV positive cohort who did not develop a HAI. This was adjusted for admission to the ICU, as an indicator for severity of infection.
Those variables statistically significant in the univariable analyses were then included in a multivariable analysis using GEE along with potential confounders; relevant models were designed for each type of infection separately (UTI, BSI or RTI), based on statistical or clinical significance. The odds ratios, 95% confidence intervals, p values along with the Quasi Likelihood under Independence model criteria (QIC) statistic were calculated for each multivariable model (Pan, 2001).
A sensitivity analysis was performed, censoring all the dead, censoring the dead who did not develop an HAI, and including all the dead in the analysis. As there was not a significant change in the estimates, the final analysis included all patients regardless of whether they died while in hospital.
The incidence density ratio of infection was calculated by dividing the number of infections by the total patient days. The time to event of infection was estimated by the Kaplan- Meier method (Kaplan and Meier, 1958).
The analyses were conducted using SAS statistical software, version 9.2 (SAS Institute, Inc., Cary, North Carolina).
Results
There were 3,877 discharges among 1,911 HIV infected patients, 2,527 (65.2%) male and 1,350 (34.8%) female discharges. The mean age was 45.93 ± 9.97 years with the youngest patient being 18 and the oldest 89. There were 282 (7.3%) discharges in which one infection occurred, 47 (1.21%) with two infections and three (0.08%) with three infections, including a total of 142 UTI, 106 BSI and 100 RTI.
Epidemiology of HAI in HIV positive patients
The incidence rate (IR) of UTI was 4.35 infections per 1,000 patient days, the IR for BSI was 3.16 per 1,000 patient days of observation and for RTI, the IR was 2.98 per 1,000 patient days. The median time to infection was around the sixth day of admission (Figure 1). Total length of stay was significantly longer for those who developed an infection compared to those who did not. Figure 2 displays the mean length of stay between those who develop a HAI compared to those who didnot develop a HAI in the HIV cohort.
Figure 1.
Kaplan-Meier Plot indicating time to infection for bloodstream infection among HIV infected patients
ǂ graphs for urinary tract infection and respiratory tract infection were similar
Figure 2.
Mean total length of stay among HIV positive discharges with and without healthcare associated infection for the different types of infections
UTI: urinary tract infection; BSI: bloodstream infection; RTI: respiratory tract infection
The causative organisms are displayed in Table 1. Among the organisms included in this analysis, the most common organisms for UTI were Enterococcus faecalis (58.1%) followed by Klebsiella pneumoniae (26.5%). The most common organisms associated with BSI were Staphylococcus aureus (48.8%) and then Enterococcus faecalis (34.2%), and for RTI, Staphylococcus aureus (56.6%) and then Pseudomonas aeruginosa (16.9%)
Table 1.
Frequency distribution of organism by site specific healthcare associated infection
| Organism | UTI n (%) | BSI n (%) | RTI n (%) |
|---|---|---|---|
| Acinetobacter baumannii | 4 (2.8) | 1 (0.9) | 2 (2) |
| Enterococcus faecalis | 43 (30.3) | 14 (13.2) | 5 (5) |
| Klebsiella pneumoniae | 22 (15.5) | 4 (3.8) | 13 (13) |
| Pseudomonas aeruginosa | 9 (6.3) | 1 (0.9) | 14 (14) |
| Staphylococcus aureus | 5 (3.5) | 20 (18.9) | 47 (47) |
| Strepotococcus pneumoniae | 0 (0.0) | 1 (0.9) | 2 (2) |
| Not of interest | 59 (41.5) | 65 (61.3) | 17 (17) |
| Total | 142 (100) | 106 (100) | 100 (100) |
Predictors of HAI in HIV positive patients
Tables 2–4 summarise the predictors of the three types of infections in univariable and multivariable analyses. The insertion of an invasive device was a significant risk factor for the development of each type of HAI. The risk ratio for developing a UTI associated with a urinary catheter was 4.1 (95% CI: 2.9–5.6) times greater compared to those without urinary catheters; the risk ratio for BSI with central venous (CV) lines was 3.8 (95% CI: 2.4–5.9) times greater compared to those without CV lines; and for those who were mechanically ventilated the risk ratio for RTI was 9.6 (95% CI 5.9–15.6) times greater compared to those who were not ventilated. Although renal failure was significantly associated with risk of each type of HAI in univariable analyses, there was still a significant association between invasive device use and HAI after adjusting for renal failure in multivariable analysis. In multivariable analysis, length of stay prior to infection was significant for UTI, but not for BSI or RTI.
Table 2.
Variables associated with healthcare associated urinary tract infection among HIV positive patient discharges. Univariable and multivariable analysis
| Variable | UTI/Number (%) | Crude odds ratio (95% CI) | Adjusted odds ratio (95% CI)** |
|---|---|---|---|
| Sex | |||
| Male | 80/2,527 (3.2) | 1 | |
| Female | 62/1,350 (4.6) | 1.52 (1.06-2.59) | 2.12 (1.28–3.51) |
| Urinary catheter | |||
| No | 66/2,920 (2.3) | 1 | |
| Yes | 76/862 (8.8) | 4.06 (2.93–5.61) | 2.54 (1.56–4.12) |
| Missing | 0/95 | ||
| Steroids | |||
| No | 78/2,808 (2.8) | 1 | |
| Yes | 64/975 (6.6) | 2.38 (1.73–3.28) | 1.75 (1.10–2.78) |
| Missing | 0/94 | ||
| CD4 ( cells/ul) | |||
| <200 | 21/654 (3.2) | 1 | |
| ≥200 | 52/1,308 (4.0) | 1.15 (0.66–1.99) | __________________ |
| Missing | 1,915 | ||
| ARV | |||
| No | 105/3140 (3.3) | 1 | |
| Yes | 37/737 (5.0) | 1.51 (0.99–2.3) | __________________ |
| TMP-SMX | |||
| No | 8/546 (1.5) | 1 | |
| Yes | 71/1,336 (5.3) | 3.49 (1.76–6.91) | 2.55 (1.22–5.34) |
| Missing | 1,995 | ||
| Length of staya | ________________ | 1.05 (1.03–1.06) | 1.03 (1.01–1.05) |
| Agea | ________________ | 1.01 (0.99–1.03) | __________________ |
Continuous variable
Adjusted for all significant univariable predictors and renal failure
ARV: antiretroviral agents; UTI: urinary tract infection; TMP-SMX: trimethoprim-sulphamethoxazole
Table 3.
Variables associated with healthcare associated bloodstream infection among HIV positive discharges. Univariable and multivariable analysis
| Variable | UTI/Number (%) | Crude odds ratio (95% CI) | Adjusted odds ratio (95%CI)** |
|---|---|---|---|
| Sex | |||
| Male | 74/2,527 (2.9) | 1 | |
| Female | 32/1,350 (2.4) | 0.79 (0.52–1.20) | _________________ |
| CV line | |||
| No | 78/3,445 (2.3) | 1 | |
| Yes | 27/341 (7.9) | 3.75 (2.37–5.92) | 1.10 (0.53–2.27) |
| Missing | 1/91 (0.01) | ||
| Steroids | |||
| No | 54/2,808 (1.9) | 1 | |
| Yes | 51/975 (5.2) | 3.76 (1.70–8.32) | 1.90 (1.08–3.32) |
| Missing | 1/94 (0.01) | ||
| CD4 (cells/ul) | |||
| <200 | 24/654 (3.7) | 1 | |
| ≥200 | 36/1,308 (2.8) | 0.73 (0.42–1.25) | _________________ |
| Missing | 1,915 | ||
| ARV | |||
| No | 78/3,140 (2.5) | 1 | |
| Yes | 28/737 (3.8) | 1.5.4 (0.99–2.40) | _________________ |
| TMP-SMX | |||
| No | 7/546(1.3) | 1 | |
| Yes | 62/1,336 (4.6) | 3.49 (1.76–6.91) | 3.19 (1.41–7.22) |
| Missing | 1995 | ||
| Length of staya | ________________ | 1.04 (1.02–1.05) | 1.01 (0.99–1.03) |
| Agea | ________________ | 1.00 (0.98–1.02) | _________________ |
Continuous variable
Adjusted for all significant univariable predictors and renal failure
ARV: antiretroviral agents; UTI: urinary tract infection; TMP-SMX: trimethoprim-sulphamethoxazole
Table 4.
Variables associated with healthcare associated respiratory tract infection among HIV positive discharges. Univariable and multivariable analysis
| Variable | UTI/Number (%) | Crude odds ratio (95% CI) | Adjusted odds ratio (95% CI)** |
|---|---|---|---|
| Sex | |||
| Male | 67/2,527 (2.7) | 1 | |
| Female | 33/1,350 (2.4) | 1.11 (0.71–1.76) | _________________ |
| Endotracheal tube | |||
| No | 72/3,579 (2.3) | 1 | |
| Yes | 28/174 (7.9) | 9.58 (5.87–15.63) | 7.06 (3.20–15.61) |
| Missing | 0/124 | ||
| Steroids | |||
| No | 42/2,808 (1.5) | 1 | |
| Yes | 58/975 (5.9) | 4.16 (2.79–6.20) | 2.53 (1.46–4.40) |
| Missing | 0/94 | ||
| CD4 (cells/µ) | |||
| <200 | 26/654 (3.8) | 1 | |
| ≥200 | 37/1,308 (2.8) | 0.69 (0.38–1.23) | _________________ |
| Missing | 1,915 | ||
| ARV | |||
| No | 78/3140 (2.5) | 1 | |
| Yes | 28/737 (3.8) | 1.19 (0.73–1.94) | __________________ |
| TMP-SMX | |||
| No | 4/546 (0.76) | 1 | |
| Yes | 64/1,336 (4.79) | 7.02 (2.46–20.01) | 5.15 (1.70–15.62) |
| Missing | 1,995 | ||
| Length of staya | ________________ | 0.99 (0.96–1.03) | _________________ |
| Agea | ________________ | 1.01 (0.99–1.03) | _________________ |
Continuous variable
Adjusted for all significant univariable predictors and renal failure
ARV: antiretroviral agents; UTI: urinary tract infection; TMP-SMX: trimethoprim-sulphamethoxazole
Neither CD4 counts nor use of antiretroviral medications were predictors of any infection. The prophylactic use of TMP-SMX, however, was a significant independent risk factor for the development of each type of HAI. The risk ratio for the UTI was 3.5 (95% CI 1.8–6.9) compared to those patients who were not on TMP-SMX, BSI risk ratio was 3.8 (95% CI 1.7–8.3) and RTI risk ratio was 7.0 (95% CI 2.5–20.0). The use of corticosteroids, either current or past, was also a significant predictor for the development of UTI, BSI and RTI.
The risk of inpatient mortality was greater among HIV-positive patients who developed a HAI compared to those who did not, however when adjusted for severity of illness using admission to ICU as an indicator of disease severity, BSI remained a significant predictor of death, even after adjusting for ICU stay. Table 5 summarises the crude and adjusted odds ratios.
Table 5.
Risk of death among HIV positive discharges with and without healthcare associated infection
| Crude odds ratio (95% CI) | Adjusted odds ratio*(95% CI) | |
|---|---|---|
| Urinary tract infection | 4.72 (2.80–7.96) | 1.81 (0.97–3.36) |
| Bloodstream infection | 8.92 (5.40–14.73) | 3.57 (1.87–6.82) |
| Respiratory tract infection | 4.26 (2.28–7.96) | 1.29 (0.66–2.50) |
Adjusted for ICU stay (yes/no) as an indicator of severity of illness
In the multivariable analyses incorporating all significant univariable predictors and adjusting for confounders in the model, different predictor variables were associated with each type of infection. This is summarised in Tables 2–4. For UTI, significant predictors included female gender, presence of urinary catheter, steroid use or TMP-SMX prior to infection and longer length of stay. The only significant predictors for BSI were steroids and TMP-SMX, and for RTI significant predictors included the presence of an endotracheal tube, use of steroids and TMP-SMX.
Discussion
Epidemiology
The incidence rates of HAI among HIV positive patients vary considerably in the literature. A study conducted in Brazil observed a rate of 8.16 infections per 1,000 patient days, (Padoveze et al., 2002) whereas a study conducted in Italy demonstrated a rate of 3.6 infections per 1,000 patient days (Petrosillo et al, 1999). A study conducted in the USA determined the incidence rate to be 11.9 infections per 1,000 patient days, which was almost double the rate of HAI as compared with other patients not infected with HIV (Goetz et al, 1994). In a multicentre prospective cohort study conducted by Stroud et al in five hospitals in the USA, an incidence rate of 6.1 per 1,000 patient days was reported (Stroud et al, 1997). In our current study, the rate of HAI in the HIV infected population was similar to that reported from Italy but was lower than previously reported in other US hospitals (Petrosillo et al, 1999). Since the comorbidities, severity of illness, and treatment factors probably varied in patients across studies, it is not possible to determine the extent to which rates are comparable
In this retrospective cohort study of HIV positive patients, UTI were the most common type of infection, followed by BSI and then RTI. Some studies have demonstrated that in HIV infected individuals BSI are more common (Stroud et al, 1997; Petrosillo et al, 1999) whereas other studies have reported that UTI are more common (Goetz et al, 1994). As with other patients, it is likely that device utilisation plays a major role in determining the frequency of organ specific infection, rather than any specific biological mechanism. The organisms responsible for infection in all the subgroups were similar to those reported in other studies and not unique to HIV positive patients (Padoveze et al, 2002; Stroud et al, 1997).
Predictors
Longer length of stay in previous studies has been shown to be a major risk factor for the development of HAI (Petrosillo et al, 1999; Weber et al, 1991; Craven et al, 1996). In this study, length of stay was predictive for UTI when adjusting for all other variables, and for BSI in the model which excluded TMP-SMX but adjusted for steroids and CV line. The reason for this is unknown; however it is possible that those patients with a BSI on TMP-SMX were sicker than the other patients and therefore when TMP-SMX was included in the model, the length of stay and CV line were no longer significant predictors. This seems unlikely, however, because we did control for ICU stay.
The fact that TMP- SMX is a risk factor for the development of a HAI as has been reported in a study evaluating nosocomial BSI infection in HIV positive patients (Petrosillo et al, 2002). We found this association in all three types of HAI, with the highest point estimate being for RTI. Long-term antimicrobial prophylaxis is known to alter the normal flora of the body, which may result in increased risk of HAI (Craven et al, 1996; Petrosillo et al, 1999).
TMP-SMX is used as a prophylactic antibiotic in HIV infected individuals to prevent opportunistic infections (Masur et al, 2002). As the drug is usually continued for several months after the CD4 count increases above 200 cells/µ, TMP-SMX may be a better marker of long-term immunosuppression compared to CD4 count in pre-hospitalized cases (Masur et al, 2002; Suthar et al, 2012). The robust association between prophylactic use of TMP-SMX and risk of HAI in this study is intriguing and suggests the need for further study.
The role of CD4 count in the literature is inconsistent. Some studies have found that a lower CD4 count (<200 cells/µ) is protective against HAI (Stroud et al, 1997), some have shown a higher CD4 count to be protective (Petrosillo et al, 1999), and others have shown no relationship between CD4 counts and risk of infection (Goetz et al, 1994). The argument for a lower CD4 count being protective is based on the premise that HIV infected individuals with CD4 counts < 200 cells/µ are on TMP-SMX prophylaxis that may confer protection against HAI (Stroud et al, 1997). This is contrary to our finding that TMP-SMX is a risk factor and CD4 count was not associated with infection in this paper. This finding is in keeping with the paper by Goetz et al (1994) and Frank et al (1997), which indicated that CD4 counts did not differ among those with and without a HAI.
Antiretroviral drugs were not a significant predictor of the risk of developing HAI. This has also been the finding in previous studies (Petrosillo et al, 1999, 2002, 2005). It would be expected that those on ARV medications would have higher CD4 counts, thus decreasing risk of infection, however this study, along with others, have shown that CD4 is not a predictor of infection (Goetz et al, 1994; Frank et al, 1997). The null finding between ARV and HAI further supports the possibility that CD4 count is not a predictor of infection and immune therapy such as ARV may have no impact in preventing HAI. In this study, HIV positive patients who developed an infection tended to stay 15–22 days longer on average in hospital, depending on the type of infection.
The inpatient mortality rate for HIV infected patients with HAI has been shown to be greater than for those HIV positive patients who do not develop a HAI, with approximately a 4–5 fold increase. In one study, the mortality rate for HIV positive patients without a HAI was 7.5% as compared with 29.7% among patients who had a HAI (Petrosillo et al, 1999). Another study indicated a 44% increased risk of death in those with a HAI (DeMarais et al, 1997). Although the risk of death was greater in HIV positive inpatients who developed a HAI in our study, rates adjusted for ICU stay were greatly attenuated. Hence, disease severity may explain a large proportion of this increased death rate. Since the study populations were different across the studies, these rates may not be comparable.
Limitations
There were several limitations of this study. It was a retrospective analysis from an existing cohort, and only data available electronically were included. The use of electronic data also limited information available to define infections. Although electronic algorithms were developed and elements from the CDC National Health Safety Network surveillance definitions for HAI were mapped to electronic data, it is likely that potential infections were missed and not included. There were considerable missing data, which may have biased the results or led to a null association. In the example of CD 4 counts, it is possible that the smaller numbers may increase the risk of a Type II error. Ideally we would have liked to assess viral load as a predictor of infection, however we did not have sufficient data. Hence, further research is needed. There may have been other confounders which were not accounted for in this study. TMP-SMX was examined in this study due to its continued use as a prophylactic agent in HIV positive patients. It may have been possible that there may have been patients who were switched from prophylaxis to treatment for Pneumocystis jirovecii pneumonia (PJP) RTI as we did not have the dose or duration of treatment. This may have resulted in reverse causation, but as TMP-SMX was associated with UTI and BSI as well, it was deemed unlikely. Other antibiotics were not assessed. The causative organisms analysed in this study were limited to the six organisms of interest in the parent study because they were the most commonly seen antibiotic resistant organisms in the three study hospitals; other aetiological agents were not included. Since the study was conducted in a large urban academic health centre, results may not be generalisable to other care settings.
Conclusion
The role of TMP-SMX and its association with the risk of HAI needs to be better understood as this could have significant health implications for the treatment of HIV positive patients, namely there being no cost effective alternative in preventing PJP and other opportunistic infections without predisposing to an increased risk of HAI. This may possibly impact on patients in the developing world due to cost factors and lack of alternatives.
Patients who are on long-term prophylaxis with TMP-SMX, such as HIV positive patients, are assumed to have a degree of immunosuppression, and therefore protective measures should be considered.
Every effort must be made to minimise invasive procedures in HIV positive patients and if there is a need for invasive lines or mechanical ventilation, it must be for the shortest duration possible. Non-invasive ventilation must be tried before resorting to invasive ventilation. In male patients, condom cathetherisation could be made mandatory in cases of non-prostatic obstruction rather than an invasive catheter. Sterility must be optimised when doing invasive procedures, and it may well be necessary to do these in a sterile setting such as theatre, however this may result in higher medical expenses and pose logistical challenges in the developing world.
Further research is required in this field in order to understand, prevent and optimally treat HIV positive patients with HAI.
Footnotes
Declaration of conflicting interest: The author declares that there is no conflict of interest.
Funding: This study was supported by the National Institute of Nursing Research (R01 NR010822) Mohammed Mitha was supported by the Columbia University-Southern African Fogarty AIDS International Training and Research Program (AITRP) through the Fogarty International Center, National Institutes of Health (grant # 5 D43 TW000231).
References
- Apte M, Neidell M, Yoko Furuya E, Caplan D, Glied S, Larson E. (2011) Using electronically available inpatient hospital data for research. Clinical and Translational Science 4: 338–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Centers for Disease and Control. (2011) HIV surveillance, United States,1981–2008. MMWR 60: 683–9. [Google Scholar]
- Craven DE, Steger KA, Hirschhorn LR. (1996) Nosocomial colonization and infection in persons infected with human immunodeficiency virus. Infection Control and Hospital Epidemiology 17: 304–18. [DOI] [PubMed] [Google Scholar]
- Demarais PL, Gertzen J, Weinstein RA. (1997) Nosocomial infections in human immunodeficiency virus-infected patients in a long-term-care setting. Clinical Infectious Diseases 25: 1230–2. [DOI] [PubMed] [Google Scholar]
- Frank U, Griffith J, Daschner FD, Schulgen G, Mills J. (1997) Incidence and epidemiology of nosocomial infections in patients infected with human immunodeficiency virus. Clinical Infectious Diseases 25: 318–20. [DOI] [PubMed] [Google Scholar]
- Goetz AM, Squier C, Wegener MM, Muder RR. (1994) Nosocomial infections in the human immunodeficiency virus-infected patient: a two-year survey. American Journal of Infection Control 22: 334–9. [DOI] [PubMed] [Google Scholar]
- Kaplan EL, Meier P. (1958) Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 53: 457–81. [Google Scholar]
- Klevens RM, Edwards Jr CLR, Horan TC, Gaynes RP, Pollock DA, Cardo DM. (2007) Estimating health care-associated infections and deaths in US hospitals, 2002. Public Health Reports (1974-) 122: 160–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Masur PBH, Kaplan JE, Holmes KK. (2002) Guidelines for preventing opportunistic infections among HIV-infected persons—2002: recommendations of the US Public Health Service and the Infectious Diseases Society of America. Annals of Internal Medicine 137: 435–78. [DOI] [PubMed] [Google Scholar]
- Padoveze MC, Trabasso PN, Branchini MLM. (2002) Nosocomial infections among HIV-positive and HIV-negative patients in a Brazilian infectious diseases unit. American Journal of Infection Control 30: 346–50. [DOI] [PubMed] [Google Scholar]
- Pan W. (2001) Akaike’s information criterion in generalized estimating equations. Biometrics 57: 120–5. [DOI] [PubMed] [Google Scholar]
- Petrosillo N, Nicastri E, Viale P. (2005) Nosocomial pulmonary infections in HIV-positive patients. Current Opinion in Pulmonary Medicine 11: 231–5. [DOI] [PubMed] [Google Scholar]
- Petrosillo N, Pugliese G, Girardi E, Pallavicini F, Carosi G, Moro ML, Ippolito G, Ospedaliere TGHEI. (1999) Nosocomial infections in HIV infected patients. AIDS 13: 599–605. [DOI] [PubMed] [Google Scholar]
- Petrosillo N, Viale P, Nicastri E, Arici C, Bombana E, Casella A, Cristini F, De Gennaro M, Dodi F, Gabbuti A, Gattuso G, Irato L, Maggi P, Pallavicini F, Pan A, Pantaleoni M, Ippolito G, Ospedaliere TGHEI. (2002) Nosocomial bloodstream infections among human immunodeficiency virus–infected patients: incidence and risk factors. Clinical Infectious Diseases 34: 677–85. [DOI] [PubMed] [Google Scholar]
- Pittet D, Allegranzi B, Storr J, Nejad SB, Dziekan G, Leotsakos A, Donaldson L. (2008) Infection control as a major World Health Organization priority for developing countries. Journal of Hospital Infection 68: 285–92. [DOI] [PubMed] [Google Scholar]
- Stroud L, Srivastava P, Culver D, Bisno A, Rimland D, Simberkoff M, Elder H, Fierer J, Martone W, Gaynes R. (1997) Nosocomial infections in HIV-infected patients: preliminary results from a multicenter surveillance system (1989–1995). Infection Control and Hospital Epidemiology 18: 479–85. [DOI] [PubMed] [Google Scholar]
- Suthar AB, Granich R, Mermin J, Van Rie A. (2012) Effect of cotrimoxazole on mortality in HIV-infected adults on antiretroviral therapy: a systematic review and meta-analysis. Bulletin of the World Health Organization 90: 128–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tumbarello M, Tacconelli E, De Gaetano Donati K, Leone F, Morace G, Cauda R, Ortona L. (1998) Nosocomial bloodstream infections in HIV-infected patients: attributable mortality and extension of hospital stay. Journal of Acquired Immune Deficiency Syndromes 19: 490–7. [DOI] [PubMed] [Google Scholar]
- Weber DJ, Becherer PR, Rutala WA, Samsa GP, Wilson MB, White II GC. (1991) Nosocomial infection rate as a function of human immunodeficiency virus type 1 status in hemophiliacs. The American Journal of Medicine 91: S206–S212. [DOI] [PubMed] [Google Scholar]