Frailty increases the risk for developing urinary tract infection among 79,887 patients with diabetic mellitus and chronic kidney disease

From the LCDP during the study period, we identified 79,887 patients with DM and CKD after applying the exclusion criteria (Fig. 1). These participants were divided into those without any FRAIL item (n = 28,832, 36.1 %), with 1 (n = 40,216, 50.3 %), 2 (n = 10,189, 12.8 %), and ≥ 3 (n = 650, 0.8 %) FRAIL items, respectively, at the beginning of this study. Those with increasing numbers of FRAIL items, or a higher frail severity, were significantly older, more likely to smoke or have alcoholism, had a higher prevalence of most comorbidities (cardiovascular, metabolic, pulmonary, hepatobiliary, oncologic, neuropsychiatric, and orthopedic ones) and were more likely to receive certain types of medications compared to those without any FRAIL item (Table 1). Participants with DM, and CKD and more FRAIL items had significantly greater severity of DM than those with DM, CKD but without any FRAIL item.

Among those with at least 1 FRAIL item, Illness was the most prevalent out of the 5 items (58 %), followed by Fatigue (17.1 %) and Loss of weight (1.3 %) (Table 2). After a mean 3.51 years of follow-up, totally 11,175 UTI events occurred, equivalent to an incidence density of 39.8 events per 1000 person-year. Kaplan-Meier event curve analysis showed that participants with DM and CKD and an increasing numbers of FRAIL items had a progressively higher incidence of UTI than those without any FRAIL item (log rank p < 0.001) (Fig. 2 A). Univariate analysis found that participants with DM and CKD and 1, 2, and ≥ 3 FRAIL items had a hazard ratio (HR) of 1.7 (95 % confidence interval [CI] 1.63–1.77), 2.03 (95 % CI 1.92–2.16), and 3.03 (95 % CI 2.57–3.58) for developing UTI than those without any items (Table 3). Cox proportional hazard modeling revealed that after accounting for all variables in Table 1, having more severe frailty was predictive of a higher risk of UTI (for groups of 1, 2, and ≥ 3 FRAIL items, HR 1.19, 1.24, and 1.43, respectively; all p < 0.001) (Table 3). An average of 11 % risk elevation for UTI was observed for every FRAIL item increase. After accounting for mortality as the competing event, having frailty at the beginning of follow-up was still associated with a significantly higher risk of UTI during follow up, with a dose-response relationship (for 1, 2, and > 2 FRAIL items, HR 1.14, 1.16, and 1.28, 95 % CI 1.08–1.21, 1.08–1.25, and 1.06–1.53, respectively) (Table 3). We found an essentially similar dose-responsive relationship between the severity of frailty and the risk of developing UTI in patients with < 65 and ≥ 65 years, male and female patients, and those with earlier stages of CKD (Table 4). However, the estimation of risk did not reach significance in those with stage 5 CKD due to a low patient number in that subcohort.

A secondary analysis focused on the risk of developing urosepsis among participants with DM and CKD according to the severity of frailty. The incidence density of urosepsis in these patients was 11.4 events per 1000 person-year. Kaplan-Meier analysis similarly showed that participants with DM, CKD, and an increasing number of FRAIL items had rising incidence of urosepsis (log rank p < 0.001; Fig. 2B) and a higher risk over time than those without any FRAIL item (Table 3). Cox proportional hazard regression showed that there was a trend of increasing urosepsis risk when participants had a higher frail severity among patients with DM and CKD (Table 3).

A sensitivity analysis incorporating individual FRAIL item positivity instead of FRAIL item counts was performed (Table 5). We found that positivity involving the Fatigue (HR 1.06, 95 % CI 1.01–1.12) or Illness (HR 1.22, 95 % CI 1.15–1.29) item was independently associated with a higher risk of developing UTI among patients with DM and CKD. In addition, positivity involving the Ambulation item (HR 1.47, 95 % CI 1.08–1.99) was independently associated with a higher risk of developing urosepsis (Table 5).

In the current study, we identified and assembled a large group of patients with DM and CKD, followed by characterizing their baseline frailty status and subsequent follow-up. We showed for the first time that having frailty could predict their risk of developing UTI and a tendency for more severe episodes during the years to come in these patients. The relationship between FRAIL scale results and the risk of UTI was not mediated solely by the Illness item, which was the most prevalent one. Judging from the fact that UTI aggravates renal outcomes in patients with CKD [35] and consumes excessive healthcare resources in patients with DM and CKD [9], it would be prudent to screen for frailty in these patients and provide them with optimal frailty-directed management in order to attenuate their risk of UTI and improve their outcomes.

Uncontrolled hyperglycemia can impair the function of polymorphonuclear cells through decreasing their mobility and chemotactic ability, as well as compromise the performance of other immunocytes including macrophages and lymphocytes [36]. Complement dysfunction and defective humoral immunity also play a role in the relatively immune-suppressive status conferred by DM and CKD [37]. On the other hand, the association between frailty and the risk of UTI has not been addressed before, but preliminary evidence from other population supports the possibility of such relationship. A recent large cohort study revealed that frailty prior to the receipt of immunosuppressive therapies significantly elevated the risk of infection in patients with inflammatory bowel disease [38]. This risk surge was independent of age, other morbidities, and medications. Another systematic review summarized that the results of frail index were able to predict the risk of contracting healthcare associated infections among medically and surgically treated patients [39]. Our findings further extend their observations that frailty begets infection, by showing that frailty confers an undue risk for a specific infection type among those who are particularly susceptible. Several reasons may be responsible for the observed relationship between frailty and the UTI risk. Patients with frailty frequently have co-existing malnutrition and other geriatric syndromes including sarcopenia, cognitive dysfunction, and/or depression [40, 41]. Prior studies showed that having low mini-nutrition assessment (MNA) scores was associated with an increased UTI risk [42]. Inadequate dietary intake may precipitate hypovitaminosis D, a potential risk factor for UTI in different populations [43]. Older adults with cognitive dysfunction, especially those who have dementia, are more likely to develop dehydration, accidental falls, and UTI compared to those without [44]. Sarcopenia frequently accompanies frailty and correlates with an increased likelihood of urine retention, catheterization and impaired mobility, all of which contribute to the risk of UTI among vulnerable population such as in-patients [45]. Finally, a recent study discovered that having frailty placed an individual at a higher risk of developing urolithiasis through multiple mechanisms including hypercalciuria and immobilization [27], and UTI can be a potential complication resulting from structural damages caused by urolithiasis. We thus provided a putative summary of the potential mechanisms responsible for the observed relationship between frailty and the UTI risk based on the existing literature in Fig. 3.

Although we discovered that frailty increased the risk of UTI in patients with DM and CKD, the association between frailty and urosepsis exhibited a dose-responsive trend only (Table 3). It is possible that the severity of frailty we identified in this cohort may differ from results using other screening instruments, since FRAIL scale has a strong flooring effect and identifies fewer patients with frailty but with higher specificity [46]. A lower degree of frailty among the categorized participants may dilute the risk of developing a more severe outcome such as urosepsis relative to UTI among the entire group of participants. In addition, the relatively lower incidence of outcome (urosepsis) may also account for the paucity of a significant association between frailty and the urosepsis risk. Indeed, a population-based registry analysis from the United Kingdom involving nearly 4 million primary care patients showed that having frailty was associated with a significantly higher risk of sepsis, in which UTI constituted the main origin [47]. We expect that expanding the size of our cohort in a future study would facilitate the detection of this association.

Phenotyping the risk factors of UTI can be important both in terms of diagnosis/treatment and prognostication, based on the European Association of Urology recommendations [12]. Since frailty may alter the risk of UTI and potentially aggravate the outcomes of those with infections, we suggest that frailty could serve as a type “E” risk factor for UTI, meaning that frailty is an extra-urogenital risk factor and worsens outcome at the same time. In this sense, strategies aiming to ameliorate frailty may favorably alter the risk of UTI in patients with DM and CKD. There are several promising approaches with potential benefits of preventing frailty onset or slowing its progression, including the enhancement of physical activity, optimization of proteinaceous food intake, avoiding potentially inappropriate medication prescription, alone or in combination as an integrated program. However, in patients with DM and CKD, nutritional intervention needs to be cautious, since overzealous protein supplementation aiming to correct frailty may take the toll of more severe metabolic disturbance and deteriorate renal function. Patients with DM and CKD are also at risk of vitamin D insufficiency, which has been shown to modify the risk of frailty [48] and serves as a potential manageable target. Judging from these findings, it is prudent to devise an individualized frailty-oriented care plan tailored to the health status of patients with DM and CKD, in order to attenuate their risk of UTI. Nonetheless, more studies are needed to support the utility of this risk-mitigation strategy.

Several medications were used less frequently in patients with higher FRAIL item counts, including angiotensin-converting enzyme inhibitors, angiotensin II receptor blockers, β-blockers, clopidogrel, fibrate, and statin, compared to those without any FRAIL item (Table 1). Indeed, treatment intensity for multiple illnesses such as malignancies [49], hypertension [50], or the accessibility to medical specialist care [51] has been found to be lower among frail patients than that among non-frail ones, despite the potentially higher prevalence of these morbidities in the former group. There are also calls for safely de-prescribing relevant medications among frail older adults to avoid polypharmacy and the associated adverse sequels [50]. This is not equivalent to under-treatment for these frail patients, since their physical condition may not allow such intensive treatments. Nonetheless, we have accounted for all variables, including the morbidity status and the associated medications, in our regression analyses. Consequently, we believe that imbalances in clinical features between groups may not influence the overall validity of our findings.

Our study has its strengths and weakness. This idea has not been tested before in the literature and our results prominently enrich the existing literature. The large number of our cohort, the extensive variables considered during analyses, and a sufficient follow-up duration all increase the validity of our findings. The strategy of defining the exposure (frailty) and outcome (UTI) variable in this study has been adopted in the existing literature [14, 15, 31, 33], with credible results obtained. However, several features of this study warrant consideration before cautiously interpreting our findings. We did not arrange image examinations for urinary tract anatomy among participants and thus could not delineate the specific structural factors through which frailty might increase the risk of UTI in these patients. Second, laboratory parameters were unavailable in our cohort as we relied on diagnostic codes for identifying events. We were unable to adjust for data such as pyuria severity, leukocyte counts, or inflammatory indices in our analyses. Nonetheless, our participants were identified from the entire population with sufficiently large numbers, and we believe that the representativeness of these patients could negate most potential imbalances in laboratory data. We did not have information about these patients’ past history of UTI, so this variable was not included in our analyses. Finally, code-based identification of frailty could have intrinsic limitations due to the possibility of under-coding and variations in patient characteristics. Those who were diagnosed with specific morbidities and coded in the database tend to have more severe illnesses. A prior study comparing the agreement between administrative database and hospital records showed a 63–70 % false negative rate for certain comorbidities [52], suggesting that under-coding might not be uncommon. Thus, it is plausible that over-estimation of effect size could occur. However, this approach of using administrative database to identify frailty can still be advantageous if a large cohort is needed for testing study hypothesis. In addition, the demonstration of a dose-responsive relationship between frailty and its effect on adverse outcomes greatly support our hypothesis (Table 3). Consequently, we believe that our findings are still valid. Nonetheless, more studies are needed.

In this study, we assembled a large cohort of patients with DM and CKD and examined their risk of developing UTI according to their frailty status at the beginning of follow-up. We were able to show that those with a higher severity of frailty, in the forms of increasing FRAIL item counts, carried a rising risk of UTI over 3.51 years of follow-up compared to those without any FRAIL items. For every FRAIL item increase, there was a 11 % increase in the risk of UTI. Based on these findings, it is suggested that the prevention or reduction in frailty severity may exert beneficial influences on the probability of future UTI risk in patients with DM and CKD.