Patient’s and physician’s awareness of kidney disease in coronary heart disease patients – a cross-sectional analysis of the German subset of the EUROASPIRE IV survey

We used data of the German sample (n = 536) of the EUROASPIRE IV survey [20]. The EUROpean Action on Secondary and Primary Prevention by Intervention to Reduce Events surveys are a multinational initiative of the European Society of Cardiology and the European Association for Cardiovascular Prevention and Rehabilitation to assess quality of secondary prevention in CHD patients across Europe [19]. The study design of the EUROASPIRE IV “hospital-arm” has been reported previously [20]. Briefly, for each participating country, a geographical region with > 0.5 million people was selected in which at least one hospital offering interventional cardiology and cardiac surgery and one or more acute hospitals admitting patients with MI and myocardial ischemia. All patients hospitalized for acute or elective treatment of CHD (coronary artery bypass grafting [CABG], percutaneous coronary intervention [PCI], acute myocardial infarction, or myocardial ischemia) were identified from the hospital’s patient management systems and invited to participate in the study. This “index” CHD-event could represent the primary diagnosis of CHD as well as any subsequent episode in previously established CHD. Subjects were eligible if they were 18–79 years old and the study visit took place between 6 and 36 months after the index hospital stay. All participants provided written informed consent. At the German study center, patients were recruited from the University Hospital Wuerzburg (Dept. of Medicine I and Department for Thoracic and Cardiovascular Surgery) and the Klinik Kitzinger Land (Dept. of Medicine). The study protocol and data-handling at the German study center were approved by the Ethics Committee of the Medical Faculty of the University of Würzburg (Vote 58/12) and the data protection officers of the University Hospital and the University of Würzburg (DS-117.605-15/12).

Information on the CHD event, risk factors, clinical measurements and laboratory values (e.g. SCr at hospital admission) were obtained by retrospective review of index hospitalization charts. During the study visit, details of CHD history (e.g. MI, CABG, PCI/stent) and information on co-morbid conditions, medication, life-style and behavior were collected in personal interviews, and standardized examinations were performed according to the EUROASPIRE IV protocol, including blood pressure, weight, height, carbon monoxide (CO) in exhaled air [22]. Serum creatinine (SCr), lipid profile and HbA1c were analyzed centrally from fasting blood samples at the National Public Health Institute, Helsinki, Finland. In addition, urinary albumin/creatinine ratio (ACR) was measured locally.

In addition to the core EUROASPIRE IV protocol, we implemented a kidney module at the German study center a few weeks after the start of enrollment, thus respective information was missing in n = 62 (11.6%) individuals. Additional information relating to kidney function during the index hospitalization, i.e. details of CKD or AKI, including dialysis requirement, reported in the discharge letter, was collected by chart review. For all German patients we were able to retrospectively collect laboratory data on SCr at hospital discharge and the maximum value of SCr during the hospital stay. For patients admitted to the University Hospital Würzburg (n = 498) the Data Warehouse of the Comprehensive Heart Failure Center [23] was utilized, e.g. for extraction of SCr values, ICD-10 codes for CKD and AKI, and OPS-codes for dialysis treatment. At the study visit, we collected data on the patient’s awareness of CKD and specialist care during personal interviews. It included the following questions: “Have you ever been told by a doctor/health care provider that your kidney function is impaired, e.g. not as good as it would be expected?”; “Have you ever been told by a doctor/health care provider that you should be seen by a specialist to have your kidney function checked?”; “Have you ever been seen by a specialist to have your kidney function checked and/or treated?”.

Kidney function was categorized into CKD-G and CKD-A stages based on estimated glomerular filtration rate (eGFR_CKD-EPI) and ACR according to KDIGO (Kidney Disease: Improving Global Outcomes) [24] (Fig. 1). Due to 95.5% missing data on ACR in the hospital records, impaired kidney function during the hospital stay was described as eGFR_CKD-EPI < 60 ml/min/1.73m² at hospital admission or at discharge or any episode of AKI. AKI during the index hospital stay was defined as SCr increase of ≥ 0.3 mg/dl within 48 h or SCr increase of 1.5–1.99× baseline SCr within 7 days (KDIGO AKI stage 1), SCr-increase of 2.0–2.9× baseline SCr (stage 2) and SCr-increase ≥ 3.0× baseline or SCr > 4 mg/dl or dialysis (stage 3) [25]. A binary variable CKD at the study visit was defined as all CKD-G stages G3a and higher and CKD stages G1A3, G2A2, G2A3 (i.e., largely preserved GFR but significant albuminuria).

During the index hospital stay the following risk factors were defined: history of heart failure (according to case history or echocardiographic findings of cardiac dysfunction at admission); cardiovascular risk factors known at admission or reported in the discharge letter as hypertension, dyslipidemia, diabetes, smoking and obesity (BMI ≥ 30 kg/m² at admission or explicitly stated in the discharge letter). Definitions of risk factors at the study visit were as follows: hypertension (blood pressure ≥ 140/90 mmHg, ≥ 140/85 mmHg in diabetes, ≥ 150/90 mmHg in patients > 80 yrs., ≥130/90 mmHg in patients with CKD [26]), diabetes (self-reported diabetes or impaired fasting glucose or impaired glucose tolerance), dyslipidemia (LDL cholesterol ≥2.5 mmol/L), obesity (body mass index [BMI] ≥30 kg/m²) and smoking (self-reported, or CO in exhaled air >10 ppm [20]).

Data are reported as proportions and median (inter-quartile range, IQR) and were compared across categories of interest (i.e. impaired kidney function during hospital stay, CKD at study visit) using Wilcoxon rank-sum test, Kruskal-Wallis test and χ² test/Fisher’s exact test, as appropriate. The trend of patient’s awareness, referral to specialist and visit at the specialist across CKD-G stages was analyzed by Cochran-Mantel-Haenszel test. Determinants of patients’ awareness of CKD (i.e. positive response to the first question of the kidney module: “Have you ever been told by a doctor/health care provider that your kidney function is impaired, e.g. not as good as it would be expected?”) and the fact that CKD or AKI were mentioned in the hospital discharge letter (physician’s awareness of kidney disease) in all participants were investigated by univariable and multivariable logistic regression, using backwards selection including variables statistically significant in univariate analysis. Results are displayed as odds ratio (OR) with respective 95% confidence intervals (CI). Analyses were performed on a complete case dataset, i.e. n = 474 with data on the kidney module. We tested the robustness of the multivariable models on a dataset with imputed missing data, using five imputations derived from the Markov Chain Monte Carlo method (SAS proc. mi). We also performed sensitivity analyses in which eGFR <50 ml/min/1.73 m² was used rather than <60 ml/min/1.73m² to increase the confidence in the patient suffering from true kidney disease by excluding patients with only a minor variation in SCr that made their eGFR values being slightly below 60 ml/min/1.73m². Analyses were performed using SAS 9.3 (SAS Institute Inc., Cary, NC, USA). Two-sided p-values <0.05 were considered statistically significant.

A total of 536 German patients were enrolled in EUROASPIRE IV (median age at the index hospital stay 67 years, 82% male). Median SCr at hospital admission for the index CHD event (i.e. first measurement in patient record) was 0.9 (IQR 0.8; 1.1) mg/dl, reflecting an eGFR_admission of 81.1 (66.2; 93.3) ml/min/1.73m² and 94 patients (17.6%) had eGFR <60 ml/min/1.73 m². At hospital discharge (i.e. last measurement in patient record), median SCr was 1.0 (0.8; 1.1) mg/dl, median eGFR_discharge was 78.7 (64.1; 91.4) ml/min and 100 (18.1%) patients had eGFR <60 ml/min/1.73m². AKI was observed in 94 (18.1%) subjects. Most AKI episodes represented a slight increase of SCr (AKI stage 1: 89.4%), while a rise in SCr of not more than exactly 0.3 mg/dl was observed in 16 patients. Two events were of AKI stage 2 (2.1%) and 8 (8.5%) were of stage 3, of which 3 had to be treated by acute hemodialysis. Any impairment in kidney function during the hospital stay (either eGFR <60 ml/min/1.73m² at admission, or at discharge or AKI) was observed in 172 patients (32.2%). Patients with impaired kidney function were older, more often had a history of heart failure, had a longer duration of CHD, and were more likely to receive CABG during the hospital stay. The hospital stay was on average also longer in CKD patients as compared to patients without renal impairment (Table 1). No differences in classic CV risk factors were observed. In the sensitivity analysis applying the lower eGFR cut-off, we found 38 (7.1%) subjects with eGFR <50 ml/min/1.73m² at hospital admission, n = 46 (8.7%) at hospital discharge and any impairment of kidney function during the hospital stay in n = 119 (22.3%) patients. We did not find any meaningful differences in the patient characteristics as compared to results using the eGFR 60 ml/min/1.73m² cut-off (detailed data not shown).

Of the 474 patients in whom data on the kidney module were available (see Methods), CKD and/or AKI were reported in prominent sections of the hospital discharge letter in 37 (7.8%) of all patients and in 32 (21.5%) of patients with impaired kidney function. While older age, length of hospital stay, diabetes and obesity lost their association in multivariable modeling, worse kidney function at hospital discharge (OR_eGFR 0.92 [95% CI 0.89; 0.94]) and more severe episodes of AKI (OR 93 [10; 848]) remained independently related to physician’s awareness of kidney disease. Moreover, in patients in whom the index event was the primary diagnosis of CHD, it was less likely (OR 0.38 [0.14; 1.00]) that impaired kidney function was mentioned in the discharge letter (Table 2). In sensitivity analyses, these findings were similar when imputing the missing values (data not shown).

Of patients admitted to the University Hospital Würzburg (n = 498) that had CKD or experienced AKI during the index hospital stay (n = 162), relevant ICD codes were applied to 117 (72.2%) patients after discharge. Correct coding was particularly observed in those patients with more severe stages of AKI (100% in stage 2, 75% in stage 3) and CKD (76.9% in stage G3a, 95.2% in G3b, 100% in G4 and 100% in G5).

The study visit was performed on average 1.8 years (1.1; 2.5) after the index hospital stay. At the study visit, 530 of the 536 German patients had SCr values available, with a median of 1.0 (0.9; 1.2) mg/dl reflecting an eGFR of 74.1 (60.0; 85.3) ml/min/1.73 m². ACR was measured in 526 subjects (98.1%), with a median of 4.9 (1.3; 15.8) mg/g. One patient was on chronic hemodialysis treatment (started already prior to the index-hospital stay). According to KDIGO classification of CKD (Fig. 1), 127 (24.4%) individuals had eGFR <60 ml/min/1.73m² and another 48 subjects (9.2%) had largely preserved eGFR but significant albuminuria (i.e. stages G1A3, G2A2, G2A3). Patients with CKD were more likely to be older, with a longer duration of CHD and a history of CABG, heart failure and peripheral artery disease (Table 3). Classic risk factors such as hypertension, diabetes, dyslipidemia and overweight/obesity were also more common in subjects with impaired kidney function, whereas smoking was less prevalent in CKD patients. Primary care regarding CHD was reported to be provided by a cardiologist in 111 (60%) patients with CKD, which on average also had spent a shorter time in education. In those patients, impaired kidney function was reported more frequently in the discharge letter of the index hospital stay. The sensitivity analysis using the eGFR <50 ml/min/1.73m² cut-off identified 65 patients (12.3%). Again, the differences in patient characteristics between of those subjects with more advanced CKD as compared to those with normal kidney function were overall very similar (detailed data not shown) to the results derived from the results presented above (Table 3).

Data on patient awareness were part of the kidney module at the German study center and were thus available in n = 474 participants (see Methods). Of those 158 subjects with CKD, 54 (34.2%) patients reported that they had been told about chronic kidney disease, 23 (14.6%) were referred to a renal specialist and 21 (13.3%) had been seen by a specialist. Even in those without overt renal dysfunction at the study visit, 19 (5.5%) reported to be aware of impaired kidney function, 11 (3.2%) were referred and 12 (3.5%) were seen by a specialist. Overall, greater proportions of awareness, referral or specialist care were observed in more advanced stages of kidney disease (p for trend <0.01), however, based on a limited number of observations (Table 4). Accordingly, reduced kidney function (OR_eGFR 0.94 [0.92; 0.96]) was also related to an increased level of patient’s awareness of CKD in multivariable logistic regression, aside from a history of heart failure (OR 1.99 [1.00; 3.97]), obesity (OR 1.97 [1.07; 3.64]), and the fact that renal impairment was reported in the index discharge letter (OR 5.51 [2.35; 12.9]) (Table 5). Similar results emerged from analyses of the imputed dataset.

Although CKD is common and associated with worse prognosis, only a small proportion of patients (<5–30%) are aware of their disease in population-based studies [14, 27], CKD cohorts [15, 28] and in CHD patients [21]. Despite public education programs e.g. in the US or the UK, only little improvement in CKD awareness could be observed [17, 29]. Early diagnosis is needed to inform patients about their disease and initiate appropriate treatment [30]. Patient education should be expected to improve adherence to medication and treatment targets [16, 31], but data in CKD are conflicting. Patients on renal replacement therapy (RRT) with adequate knowledge about their disease and treatment targets including dietary restrictions have a lower mortality risk when compared to less educated patients [32]. In contrast, in earlier stages of CKD, achievement of treatment targets for adequate blood pressure control was not significantly associated with the level of patient’s awareness of CKD [17]. Yet, recent data are encouraging that focused education of the primary care physician (PCP) and the patient including his relatives can indeed improve risk factor control and also slow the progression rate of CKD [33]. As directly from the patent’s perspective, it seems intuitive that one feels more confident if he is aware of a certain condition and is informed and educated accordingly about the treatment options. On the other hand, he might also be frightened by the information on the diagnosis and the disease’s prognosis [31], which may aggravate depressive mood, since depression is a common comorbidity in CHD [34] as well as in CKD [35]. Specialist care, e.g. provided by a nephrologist, may help to adequately inform patients about their individual risk for CKD progression [13], and thus help individualizing the therapeutic strategy and treatment targets. It has been shown that (early) referral to nephrology care can slow CKD progression and is associated with reduced mortality risk once RRT is initiated [36, 37].

In our sample of stable CHD patients, we found 24% with CKD (i.e., applying the commonly used cut-off of eGFR <60 ml/min/1.73m²), which is nearly 10-times higher than in recent numbers of the German general population [38]. Another 10% of our sample had albuminuria with preserved eGFR representing those at risk for CKD progression [24, 39]. Of those with CKD, only about a third were aware of their disease and only a minority was being seen by nephrologists, however, with higher likelihood of CKD awareness and specialist care in more severe stages of CKD. The latter observation, however, is based on very few data in the respective categories and need to be interpreted with caution. We could not find significant associations with CKD awareness with the level of education, gender, or diabetic status. Yet, a history of heart failure was related to a higher level of CKD awareness, independently of the severity of CKD. This might be explained by frequent appointments specifically for heart failure and potentially also for cardio-renal syndrome [40] at the cardiologist and the PCP, with an increased likelihood of impaired kidney function being detected, mentioned and discussed during such appointments. Furthermore, we found that patients were more likely to know about impaired kidney function if either CKD or AKI was mentioned in a recent discharge letter. Both factors underline the relationship of patient information as being directly dependent on the physician’s awareness, information and education [41].

Patients with kidney disease are also at a higher risk for complications in hospital stays for CHD and impaired prognosis after discharge [42, 43]. Causes for affected kidney function are multifactorial, including acutely reduced renal perfusion in myocardial infarction, nephrotoxic contrast application during catheter interventions, but also in elective CABG surgery [25]. These AKI episodes, which might be only temporary changes in kidney function of milder degree (e.g., a rise in SCr by 0.3 mg/dl in AKI stage 1 [25]) are associated with an increased risk for cardiovascular events, cardiac dysfunction, heart failure progression, and risk of hospitalization and death [44].

In general, markers of kidney function at hospital admission and during the early phase of a hospital stay are more likely to be influenced by AKI, whereas kidney function might be improved and stabilized at hospital discharge. While in the core protocol of EUROASPIRE IV only SCr values at hospital admission were collected, at the German study center, we used all SCr measurements during the index hospital stay, thus enabling detailed analysis of the course of kidney function including AKI episodes. Of note, even if all methods were considered (e.g. urinary dip-stick, 24 h urine collection, ACR or total protein/creatinine ratio), in more than half of the patients no measure of proteinuria was available. Therefore, estimating the entire spectrum of kidney disease including proteinuria was impossible.

We found that during the index hospital stay, about 18% of patients experienced AKI, while most episodes were of stage 1, but more severe stages including those needing hemodialysis were observed. Only a very small number of patients had a rise of SCr of exactly 0.3 mg/dl that did not further increase in subsequent laboratory measurements. Also about 18% were found for impaired kidney function at hospital admission and at hospital discharge, respectively. Yet, one third of all patients was detected as having any impairment of kidney function either acute and/or chronic, respectively.

Since the introduction of equations based on SCr to estimating GFR, eGFR is increasingly displayed on routine laboratory reports with every SCr measurement [45], including the recruiting German EUROASPIRE IV centers. Therefore, information on kidney function is routinely visible to the treating physician. Moreover, the role of CKD and AKI as an important comorbid condition is widely discussed in the medical literature. The rationale for using mentioning of CKD or AKI in prominent parts of the hospital discharge letter as a proxy for physician’s awareness of impaired kidney function was as follows: First, even slight changes in SCr need to be recognized by the physician as AKI, an important acute complication during the hospital stay. Second, either CKD or AKI reflect important risk factors for both CHD and CKD progression. Since the discharge letter represents the most important document of information transfer from the hospital to the ambulatory setting, the treating physician needs to judge kidney dysfunction as important enough to be clearly reported the discharge letter. In clinical routine in Germany, particularly the first part (diagnoses) and the end of the document (summary and medication) are predominantly being read by PCPs due to time constraints.

We found that the discharge letter reported only one fifth of patients with impaired kidney function (acute or chronic). While comorbid conditions, CHD history or the procedure itself were unrelated to physician’s awareness, it was reassuring that higher stages of CKD or AKI increased chances for kidney dysfunction being reported in the discharge letter. Importantly, in patients in whom the index event was the primary diagnosis of CHD, impaired renal function was frequently not reported in the discharge letter. In particular in these patients, comprehensive description of traditional and non-traditional CV risk factors is needed for establishing an individualized treatment concept for optimal secondary CHD prevention [46]. The discharge letter not only addresses the PCP, but also constitutes an important source for information to the patient himself, supporting self-empowerment, self-management and self-monitoring. Since reporting CKD or AKI in the discharge letter relates to patients’ awareness of CKD, raising the physician’s awareness of kidney disease may ultimately lead to better informed patients.

In 2003/2004, the German Diagnosis Related Group (G-DRG) system replaced the cost-based reimbursement of hospital stays employing ICD diagnoses, procedures, and comorbidities [11]. For each case, coding usually gets completed a few days after discharge, commonly with the help of expert coding assistants. As AKI and CKD increase the amount of reimbursement, adequate coding is highly relevant for the hospital. In our study, CKD and AKI were correctly coded for the majority of patients, in particular in those with advanced stages. However, there were still patients in whom adequate coding of renal function would have increased monetary benefits for the hospital.

The unique setting of the EUROASPIRE IV study including the kidney module at the German study center allows a comprehensive view on CKD awareness among CHD patients and their treating physicians, however, limitations need to be mentioned. First, the study-sample cannot be claimed as representative neither for CHD patients in Germany nor for those admitted for CHD at the recruiting hospitals due to the selection process of centers and the recruitment success [20]. Second, CKD at the EUROASPIRE IV study visit was classified based on a single measurement of SCr and ACR, while usually two independent measurements for adequate assessment of CKD are desired [24]. However, study participants were in apparently stable condition at the time of recruitment. In contrast, as also discussed above, during the index hospital stay, kidney function at admission as well as at discharge might be influenced by acute clinical circumstances, e.g. acute kidney injury at admission or the (prolonged) convalescence of kidney function after AKI. Third, any SCr-based estimation of GFR has limitations, and mild to moderately impaired kidney function may be better described by equations based on Cystatin C [47], which unfortunately was not available. However, we chose the SCr-based CKD-EPI formula as it outperforms the MDRD formula in particular in GFR between 60 and 90 ml/min/1.73m² [48]. Fourth, the patient’s knowledge of having been told about CKD may by influenced by multiple factors, including the setting of a study visit, therefore recall bias may apply. In addition, particularly in older subjects, a slight deterioration of kidney function might be considered as normal, age-related decline in GFR. Fifth, the number of variables tested in logistic regression analyses may be considered as too high, thus finding and missing associations by chance and limited power is surely possible. Finally, the substantial discrepancy between correct ICD-coding and presence of impaired kidney function in the discharge letter might be explained the fact that the discharge letter is predominantly prepared by junior physicians and may not reflect awareness by the more senior physician involved in care decisions, whereas the coding is supported by professional coders who may detect comorbid conditions that may have not been in the primary focus of the patient’s therapeutic care.