Infection-related hospitalizations over 30 years of follow-up in patients starting renal replacement therapy at pediatric age

Background Pediatric renal replacement therapy (RRT) patients surviving long-term are at a much higher risk of mortality compared with the age-matched general population. Recently, we demonstrated a transition from cardiovascular disease to infection as the main cause of death in a long-term follow-up study of pediatric RRT. Here, we explore the burden of infections requiring hospitalization over 30 years of follow-up on RRT. Methods The cohort comprised all 234 Dutch patients on RRT under 15 years of age between 1972 and1992. We analyzed infection-related hospitalizations during the period 1980–2010. We evaluated the Hospital Admission Rate (HAR) per patient-years (py) and infectious over non-infectious HAR ratio (HARR). Results The HAR decreased significantly over time for all patients. The rate of hemodialysis-related infections decreased between 1980 and 1999, but stabilized during 2000–2010, whereas peritoneal dialysis-related infections decreased progressively. Transplantation-related infections did not change, except for urinary tract infections (UTIs), which increased significantly from 3.3/100 py [95%CI 3.2–3.4] in 1980–1989 to 4.4/100 py [4.2–4.5] in 2000–2010 (p <0.001). The contribution of infection to HAR increased significantly in transplanted patients (HARR: 1980–1989: 0.25 [0.2–0.3]; 2000–2010: 1.0 [0.79–1.27], p <0.001). Conclusions Our findings indicate a relative increase in infections requiring hospitalization over time in patients starting RRT during the pediatric age, especially severe UTIs in transplantation. More attention paid to urological abnormalities in cases of recurrent UTI and tailored adjustment of immunosuppression may reduce risk in these patients.


Introduction
Patients starting renal replacement therapy (RRT) during the pediatric age and surviving long term are at a much higher risk of mortality compared with healthy peers, up to 30 times greater, according to long-term outcome studies. By definition, these data concern historical patients who underwent a different approach compared with current practices. However, although some data show a decline in death rates among end stage renal disease (ESRD) children over time, a recent paper on ESPN/ ERA-EDTA data showed a 55-fold increased risk in patients with pediatric ESRD compared with the general population [1][2][3][4][5].
Infection is one of the major causes of mortality and morbidity in patients on RRT, accounting for the 15-23 % of deaths in this population [6][7][8]. Various factors may contribute to the high rate of life-threatening infections in these patients, such as an impaired immune function as a result of decreased renal function, the open connection of the peritoneal cavity in patients on peritoneal dialysis (PD) and of the central venous system in hemodialysis (HD) patients, and, above all, the use of immunosuppressive therapy in transplanted patients [9][10][11][12][13][14]. It is of major concern that recent reports show an increasing trend in infection-related mortality, both in dialysis and transplanted patients, and not only in incident patients, but also in patients with a long history of RRT, years after transplantation [15,16], with an infectious mortality rate increasing from 0.51/ 100 patient-years (py) before 1989 to 0.82/100 py after 2000 in a Dutch cohort [17]. Infection is one of the most frequent causes of hospitalization in RRT patients [1,9,12,[18][19][20][21][22][23]. In adults it has been associated with a 10 % death rate within 30 days of admission [16]. Still, studies reporting the burden of severe nonfatal infections in long-term RRT patients are lacking.
Our primary aim was therefore to evaluate the burden of infections requiring hospitalization in patients starting RRT during the pediatric age with a 30-year follow-up. Our secondary aim was to analyze the change in burden of severe infections over time using infection-related hospitalization as a marker. Lastly, as hospitalization patterns in the Netherlands have adopted a higher threshold for admission over the last 30 years, we also analyzed the infection/non-infection admission rate ratio over time, to correct for this change in admission policy.

Data collection
We gathered information on patients from the Late Effects of Renal Insufficiency Cohort (LERIC) study, which comprised all Dutch patients who started chronic RRT between 1972 and 1992 at less than 15 years of age, and who were born before 1979. From the original LERIC cohort, we included all patients alive on 1 January 1980 (Fig. 1). Data collection details and results of the first follow-up studies conducted on this cohort between 1998 and 2000 have been described previously [1,17]. In 2000 and 2010, coworkers from the LERIC study visited all 37 hospitals that had been involved in the medical care of patients during the observation period and collected data on, amongst other things, age at RRT start, start and end dates of HD, PD and transplantation, date and cause of death, and the cause of each hospital admission. In living patients, the day of review was considered to be the end of the observation period.

Hospital admissions
Information on any hospital admission between 1 January 1980 and the end of the observation period were collected. Causes of hospitalization were classified independently by three reviewers (JLV, JWG, and KJJ), using a detailed description of each patient status around the admission date. Admissions for child birth, dialysis access placement and transplantation were classified as Bplanned^and excluded from the analyses. Infection-related hospitalizations were classified as: airway infection, gastroenteritis, peritonitis, sepsis, urinary tract infection (UTI), vascular access infection or other infections. Other infections include all infections of unknown origin. Furthermore, bacterial infections were defined as any positive bacterial cultures, abscesses, UTIs, sepsis, tunnel/ central venous line infections, PD peritonitis (except when indicated other), sinusitis, segmental pneumonia, bilateral pneumonia with severe disease, pneumonia successfully treated with antibiotics, Bfever with chills and antibiotic treatment.^Infections defined as viral included all diagnoses due to a specific virus, gastroenteritis except when indicated as bacterial, and upper airway infection.

Statistical analysis
In this paper, we compare the rate of hospital admission and infection-related admission during 2000-2010 with the previous two decades (1980-1989 and 1990-1999). The Hospital Admission Rate (HAR) was calculated as the number of hospital admissions per 100 py on RRT with a 95 % confidence interval (95 % CI). Hospital Admission Rate Ratios (HARRs) were calculated to compare infection and non-infection-related admissions. Poisson regression models were used to examine the trends in infection and non-infection HAR with the decade of admission (1980-1989, 1990-1999, and 2000-2010). In the Poisson model, the natural logarithm of the total py at risk was used as the offset. Data analysis was performed using R (v. 3.0.1, R Foundation for Statistical Computing) [24].

Study cohort
The total cohort consisted of 234 patients with a median follow-up of 25.0 years (range 0.2-31.4). The median age at the start of RRT was 11.2 years (range 1.9-15.0), and 55.6 % were female. The patient characteristics for each decade are shown in Table 1

Infection-related hospital admission rates
Infection-related HAR decreased between 1980-1989 and 1990-1999, but stabilized in 2000-2010. Infection-related HAR in HD patients also decreased from 1980-1989 to 1990-1999, and stabilized after 2000, whereas infectionrelated HAR in PD patients showed a progressive decrease over time. Conversely, infection-related HAR in transplant patients did not change significantly during the study period, with an almost constant HAR from 11.9 during the 1980s to 9.9 out of 100 py during the 1990s and 2000s.
During the first year after transplantation, infection-related HAR was consistently more frequent compared with the period after the first year. Furthermore, there was no improvement over time for either the first-year risk, or the long-term risk of infection-related admission, with a HAR in the 1980s of 30.7 (95 % CI: 29.8-31.5) for the first year vs 8.5 (95 % CI: 8.4-8.7, p <0.0001) for later infections and 34.1 (95 % CI: 32.5-35.8) vs 9.0 (95 % CI: 8.9-9.2, p <0.0001) during the 2000s (Fig. 3).

Diagnosis-specific, infection-related hospital admission rates
We classified infection-related admissions into specific diagnosis categories: airway infection, gastroenteritis, peritonitis, sepsis, UTI, vascular access infections, and other Infections. Of the 178 HD-related admissions due to infection, 30 (16.8 %) were airway infections, 28 (15.7 %) sepsis, 26 (14.6 %) vascular access infections, and 94 were other types of infection. Rates of airway infection, sepsis, and other infections did not show any significant trend between 1980 and 2010; however, nearly all vascular access infections (21, 11.8 %) occurred during the 1980s (Fig. 5a).
Of the 155 PD-related admissions due to infection, peritonitis was by far the most common cause of admission (111, 71.6 %). The peritonitis-specific admission rate remained constant throughout the follow-up period (Fig. 5b); however, we found a significant increase in peritonitis/other infection ratio from the 1980s (1.4, 95 % CI: 1.3-1.5) to the later decades (3.2, 95 % CI: 3.0-3.5 in the 2000s).

Discussion
We analyzed the burden of severe infections over 30 years of RRT in patients with pediatric onset of RRT using hospitalizations as a marker. We found a consistent increase in the Very few data exist on the long-term effects of RRT, as most studies report on data covering much shorter periods of observation. Our data show a high burden of transplant-related infections, not only during the first year after transplantation, as previously described by others [25][26][27][28], but also long afterward. We found no significant changes in the first year/later infection ratio throughout the observation period. This is especially remarkable as these patients were on average between 30 and 40 years old in 2000-2010, an age that is associated with the lowest risk of death by infection according to most RRT registry studies [29,30]. Our findings, therefore, most probably reflect a more general trend toward a relative increase in infections in renal transplant patients over the last 10 years. This trend is in line with the shift from cardiovascular to infectious disease as the primary cause of death, on which we previously reported [17], as well as with reports on admission rates for infections during the first posttransplant year [15,25,29,31,32].
Infections, even if not lethal, are especially worrying in transplantation patients as they may lead to graft loss. UNOS data have shown an increase in death-censored graft failures due to infections between 1997 and 2006 from 6.4 to 10.1 % [33]. Infection may lead to graft failure in several ways. It may activate the immune system and trigger cytokines that may induce interstitial inflammation, leading to chronic allograft nephropathy. BK and CMV viruses may cause tubulointerstitial nephritis and bacterial pyelonephritis, and may also directly damage the renal graft [21]. According to the UNOS data, UTIs related to urological complications were associated with an 8.8-times increased risk of death-censored graft failure [33].
The most plausible cause for the increase in burden of transplant-related infections is the concurrent tendency toward the use of more potent immunosuppressive strategies in renal transplantation over the last 20 years [34]. There is abundant evidence for a direct relationship between the extent of immunosuppression after transplantation and the risk of infection [31,35,36]. It is not the specific type of drug, but the use of higher dosages, especially of calcineurin inhibitors, and the use of triple instead of double therapy, that have been found to be associated with both a decrease in acute rejections and, as a tradeoff, more infections, such as polyomavirus and UTI [37][38][39][40]. In the Netherlands, and similarly in many other countries, all centers for adult renal transplantation introduced IL-2 blockers during the early 2000s as part of induction therapy, and all centers have switched from cyclosporine to high-dose tacrolimus in combination with mycophenolate mofetil as part of induction therapy. Indeed, this implies a substantial increase in the average immunosuppressive dose over the last 10 years [34]. A Spanish observational study showed that basiliximab as part of induction therapy was associated with UTI after renal transplantation [28]. The observed rise in hospital admissions related to UTI in transplantation patients may also be due to the increasing use of antibiotic prophylaxis after renal transplantation, causing a rise in multi-resistant infections that frequently require intravenous antibiotic treatment and hospitalization [40,41]. In addition, a general increase in the incidence of infections in Western countries has been reported, possibly due to several factors, such as an increase in multi-resistant infection incidence and improved detection and reporting practices [35,42,43].
For PD, we observed a gradual decrease over time in infection-related admissions, except for peritonitis. The persisting burden of PD-related peritonitis contrasts with data from two European centers, an Australian center and a Canadian Registry, all showing a decreasing trend over time [44][45][46][47]. An explanation might be that these studies analyzed different patients in different eras in contrast to ours, where we followed one closed cohort of patients over time. In these studies, profiles of patients within the different eras did not vary significantly in duration of RRT and, consequently, comorbidity, whereas in our study, patients who were on PD in the decade 2000-2010 had, by definition, more years of RRT and a greater dialysis burden than in the years before. Conversely, those surviving to 2000-2010 may be considered less susceptible to various peritonitis risk factors as they can be considered Bsurvivors^. Therefore, this finding raises concern, especially as it is also at odds with the major reduction in vascular access-related infections seen in HD over time.
This study has several limitations. First, we used hospitalization as a marker for the burden of severe infections. Hospitalization has been used as an indicator of disease severity for various other diseases, as severity assessment scores may have even more limitations [48][49][50]. Changes in hospitalization rates over time are fraught with difficulties because they may be subject to underlying trends in patterns of care. A change in the rate of hospitalizations not only reflects disease severity or incidence, but also organizational issues, such as improved methods of outpatient care, a change in the diagnostic criteria for infections, and changes in nephrologist hospitalization and treatment policies from the 1980s up to the present day. In Europe, there is a general trend toward lower numbers and shorter hospital stays. This is clearly reflected in the decreasing number of non-infectious-related hospitalizations. Therefore, the relatively constant rate of infectious hospitalizations may even be flawed, as in reality there may have been a sharp increase in the number of infections, but patients were not admitted to hospital. Another limitation is the retrospective collection of the data, which in some cases hampered an accurate distinction between bacterial and other causes of infection. Finally, following a closed cohort of patients over such a long period, patients surviving up to 2000 after a long history of RRT may represent a very specific population that may be difficult to compare with all prevalent RRT patients. For instance, patients returning to dialysis after transplant failure could be at a higher risk of infection than naïve dialysis patients.
In short, we found evidence for a significant increase in the burden of clinically significant infections in transplanted patients over the past decades, not only in the first year after transplantation, but also among patients who have been living with a functioning graft for a prolonged period of time. This high risk of infection should be taken into account in the management of patients with a long history of RRT, such as a more tailored down immunosuppressive regimen in patients with no history of rejection and the specific analysis of urological function in patients with recurrent UTI. This study was performed as part of the LERIC follow-up study, which is mainly funded by the Dutch Kidney Foundation. The funder played no role in the design and conduct of the study, in the data gathering or interpretation, in the preparation of the manuscript, or in the decision to submit the report for publication.
Contributions Research idea and study design: JG; data acquisition: JV and JG; data analysis/interpretation: DL, JV, KvS, KJ, and JG; statistical analysis: DL; supervision or mentorship: KJ and JG. Each author contributed important intellectual content during manuscript drafting or revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved. JG takes responsibility that this study has been reported honestly, accurately, and transparently; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.
Ethical disclosure We obtained permission from the medical ethics committee and informed consent from all patients was obtained at the first and second waves of data collection in 2000 and 2010.

Conflict of interest The authors declare no conflict of interest.
Open Access This article is distributed under the terms of the Creative Comm ons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.