Poor recognition and undertreatment of anemia in patients with chronic kidney disease managed in primary care

Anemia is a common complication of chronic kidney disease (CKD), but limited awareness and treatment options may hinder its management among CKD patients followed in primary care.


Introduction
Approximately 10% of adults are affected by chronic kidney disease (CKD) globally [1], which is a condition characterized by reduced kidney function or signs of kidney damage that are sustained over time [2].CKD is disproportionally present in adults with advanced age (prevalent in >40% of people over 60 years of age) and in those with hypertension, diabetes, and cardiovascular disease [3,4].CKD is associated with considerable comorbidity, which results in increased health care utilization, treatment burden, and mortality [5].
Given the size of the population affected by CKD and that many patients with CKD are at low risk of progression to kidney failure [6], primary care physicians (PCPs) represent the front line in the early identification and management of CKD [7].Management by nephrologists typically occurs at advanced CKD (eGFR <30 mL/min/1.73m 2 ) [8], but geographically diverse studies consistently report that most adults with CKD who are managed in primary care remain undetected, undiagnosed, and undertreated [9][10][11].In patients with a first-detected eGFR indicative of "probable" CKD (eGFR <60 mL/min/1.73m 2 ) in Stockholm's primary health care, only 3%-7% received a clinical diagnosis of CKD at baseline, and about half of these patients did not receive treatment with reninangiotensin inhibitors (RASi) [12].In the following 18 months, 28%-34% had an albuminuria measurement, and 8%-13% met referral criteria for visiting a nephrologist [12].Barriers that hinder effective CKD care by PCPs may include a lack of awareness and/or understanding of guidelines for risk stratification and management of CKD, lack of confidence in managing CKD, and limited communication channels with specialists in nephrology [7].Additionally, PCPs may have limited time to manage complex visit agendas [7].
Anemia is one of the common complications of CKD, resulting from the impaired synthesis of erythropoietin by the kidneys, due in part to impaired oxygen sensing, dysregulated iron metabolism, and decreased survival of red blood cells [13][14][15].Anemia of CKD is a risk factor for cardiovascular morbidity, mortality, and reduced quality of life [13,16] and may already manifest at eGFR <60 mL/min/1.73m 2 [17].Although the management of anemia in CKD has been studied in nephrologist care settings [18], anemia is not well studied in people with CKD managed in primary care.Cross-sectional analyses using data from primary care facilities in the UK [19] and a large US health database [20] reported suboptimal management of anemia in patients with non-dialysis-dependent (NDD)-CKD.These studies revealed infrequent testing of ferritin, low use of oral iron, and an inappropriate prescription of concurrent medications that may exacerbate anemia.It is unknown whether such clinical gaps exist in other regions.Moreover, the preceding evidence is limited by its cross-sectional nature, therefore only evaluating completed actions across survivors or insured participants, and unable to distinguish decisions taken by the PCP alone from those taken via referral to the nephrologist.
In this observational study with prospective data collection, we evaluated the incidence of anemia and subsequent processes of care in patients with NDD-CKD stages 3-5 who were managed in primary care in Stockholm, Sweden from 2012 to 2018.

Data source
We used data from the Stockholm CREAtinine Measurement (SCREAM) project, a repository of laboratory information gathered during health care interactions with primary, secondary, and tertiary caregivers in the complete population of Stockholm, Sweden [21].A single health care provider in the Stockholm region provides universal and tax-funded health care to 20%-25% of the population of Sweden [21].Using each citizen's unique personal identification number, laboratory tests were linked to other supplementary databases (regional and national) containing information on demographics, health care use, dispensed drugs, validated renal outcomes, diagnoses, and vital status, with virtually no loss to follow-up [22].The Regional Ethical Review Board in Stockholm approved the study, and informed patient consent was not deemed necessary since all data were deidentified after linkage of the databases.

Study population
For this study, we included adults aged ≥18 years from the SCREAM cohort residing and accessing health care in Stockholm from 2012 to 2018.Patients were included in outpatient care at the first eGFR measurement <60 mL/min/1.73m 2 , denoting CKD stages 3-5.Using all available preceding eGFR measurements, we then estimated if patients fulfilled the chronicity criteria by evaluating whether the interpolated eGFR remained <60 mL/min/1.73m 2 after 3-6 months [23].In these patients, we extracted data for all subsequent hemoglobin (Hb) tests.Real-world health care differs from controlled settings, with Hb testing performed based on the judgment of the attending physician and an underlying indication.To minimize confounding by indication bias, we excluded Hb tests performed within 30 days after a bleeding event or a transfusion code, or Hb tests taken during an inpatient stay or emergency room visit (±1 day).We also excluded Hb tests performed within 30 days from a hospitalization discharge, which could relate to the monitoring and/or resolution of an event.After performing these exclusions, we selected the date of the first Hb test per patient as the index date for our study and the timepoint where baseline covariates were assessed and then follow up began.Once the potential study population had been formed, we excluded individuals with anemia at index date (defined as an anemia diagnosis in the year prior), and those with a history of visiting a nephrology department within the previous 2 years.We also excluded individuals who had conditions affecting the interpretation of Hb values, which were patients who: had a recent pregnancy or experienced childbirth; had an ongoing or recent history of cancer, hematologic disease, or leukemia within the previous 3 years; or had a chronic infection (e.g., hepatitis, tuberculosis, or human immunodeficiency virus).

Exposures and outcomes
This study comprised two analyses.For the first analysis, the patient cohort was followed for the development of incident anemia (primary outcome).We considered only the first-detected anemia event per patient, defined as a low Hb measurement (World Health Organization [WHO] definition: <12 g/dL for females or <13 g/dL for males [24]), followed by a diagnosis of anemia (International Classification of Diseases 10th Revision [ICD-10] codes D60-64; Table S1) or treatment initiation within 3 months, or a subsequent Hb measurement with similar magnitude between 3 and 6 months apart (i.e., anemia sustained for at least 3 months).We also defined anemia events by severity, for which severe anemias were defined as a very low Hb measurement (<10 g/dL regardless of sex) followed by a diagnosis of anemia, treatment initiation, or a second Hb measurement <10 g/dL.Mild/moderate anemia was defined as an Hb measurement <12 g/dL for females or <13 g/dL for males but ≥10 g/dL, followed by a diagnosis of anemia, treatment initiation, or a second Hb measurement of similar magnitude.Outcomes based on anemia severity were not mutually exclusive, and a patient may have developed more than one anemia event of varying severity during follow-up.Patients were censored during follow up at the time of incident referral to nephrology care to explore the processes of care surrounding anemias in primary care.Baseline predictors of incident anemia were also identified.
In the second analysis, we investigated the association between incident anemia and the subsequent risk of major adverse cardiovascular events (MACE) and death (Table S2).Thus, anemia became a time-varying exposure.MACE was defined as a composite of nonfatal stroke, nonfatal myocardial infarction, nonfatal heart failure, and cardiovascular death.Deaths were ascertained by linkage with the Swedish population register.Anemias occurred while patients were managed in primary care settings, but we did not censor patients at nephrology referral for the evaluation of associated risks.In other words, developing anemia may have resulted in a referral to nephrology and been complicated with an adverse clinical outcome.

Clinical responses to incident anemia
We evaluated the processes of care around anemia events focusing on three different aspects: (a) recognition, defined by a clinical diagnosis of anemia within 6 months from the anemia event; (b) testing of iron parameters, defined by at least one laboratory test of ferritin or transferrin saturation (TSAT) within 6 months from the anemia event; and (c) initiation of treatments, defined by blood transfusions, infusions of intravenous (IV) iron, or filled prescriptions of oral iron or ESAs within 6 months from the anemia event.

Study covariates
Confounders were selected based on biological plausibility and included: age, sex, highest level of education, calendar year, eGFR, history of comorbidities (including diabetes mellitus, hypertension, myocardial infarction, stroke, peripheral vascular disease, heart failure, or atrial fibrillation), and current use of medications (including angiotensinconverting enzyme [ACE] inhibitors/angiotensin II receptor blockers [ARBs], beta-blockers, calcium blockers, and statins).Tables S1 and S3 provide detailed definitions of all study covariates.Study covariates were derived at index date and updated at the time of anemia development.

Statistical analysis
Values are expressed as mean and standard deviation (SD) for continuous variables with normal distribution, median with interquartile range (IQR) for variables with non-normal distribution, and number with percentage of the total for categorical variables.No missingness was identified from the records in relation to medication use, demographics, or comorbidities.
We first calculated anemia incidence rates as the number of events divided by the person-time, following patients until the first anemia event detected.We assessed time to anemia event, as well as baseline predictors, through multivariable Cox regression models, reported as hazard ratios (HRs) and 95% confidence intervals (CIs).The baseline predictors considered included: CKD stage, demographic characteristics, comorbidities, use of medications, and laboratory tests, as detailed in Table S4.For these analyses, patients were followed until the occurrence of anemia, referral to nephrologists, death, emigration from Stockholm, or end of followup.Continuous variables were standardized as per SD increase, and the relative importance of each predictor was assessed by the estimated explained variance of the outcomes (R 2 ) and the proportion of overall explainable log-likelihood (Χ 2 ) attributable to each predictor in the analysis of variance.
Next, we described clinical responses to anemia events, overall and by anemia severity, by reporting the proportion of anemia cases fitting the criteria.Finally, we estimated the associations between anemia and subsequent risk of MACE and death, which were explored through multivariable adjusted time-dependent Cox proportional hazards regression.As sensitivity analyses, we evaluated the risk of unmeasured confounding by the Evalue methodology [25], which quantifies the minimum strength of association that an unmeasured confounder would need to have with both treatment and outcome, conditional on the measured covariates, to fully explain the observed association [25].We reanalyzed associations with MACE and death after excluding events within the first 180 days after incident anemia to assess the impact of reverse causation bias (e.g., that anemia may have been identified as a result of clinical investigations following suspicion of an adverse event).Analyses were conducted in R software, version 3.6.3.

Baseline patient characteristics
After applying inclusion and exclusion criteria, we identified 62,849 patients with sustained CKD stages 3-5 undergoing routine Hb testing in primary care (Fig. S1).Of these, 17,212 (27%) patients had anemia at cohort entry and were further excluded, leaving 45,637 adult patients for our study.

Incidence of anemias
During a median follow-up of 2.4 (IQR 1.0-4.7)years, 11,987 (26%) patients developed anemia before any recorded visit to nephrology care.Most anemia cases (n = 10,339; 86%) were identified by persistently low Hb values, whereas only a small proportion were identified by anemia diagnoses (n = 917; 8%) or initiation of anemia treatments (n = 731; 6%).Cases of severe anemia were reported in 4730 (10.4%) patients.Patients developing anemia experienced a more rapid eGFR decline (on average 0.56 mL/min/1.73m 2 /year; Fig. S2) and consequently a more accentuated reduction of Hb over time (Fig. S3) compared with those who did not develop anemia.
The incidence of anemia cases, overall and by severity, increased with more severe CKD stages; for example, the incidence (cases per 1000 personyears) of anemia was 78.4 among patients with CKD stage 3a and 185.1 in those with CKD stages 4-5 (Table S5).Patients who developed anemia were most often male, with advanced age (≥80 years), and a history of diabetes, myocardial infarction, or heart failure (Table S5).Older age, male sex, lower eGFR, and a history of diabetes mellitus or heart failure were the factors with the highest relative contribution to the prediction of anemia in multivariable Cox regression (Fig. 1).

Clinical recognition and responses to incident anemia
Table 2 describes the clinical responses to incident anemia, overall and by anemia severity, within 6 months from the anemia event.The proportion of anemia cases that received a clinical diagnosis was 15.0%.The majority (67.7%) of, but not all, severe anemia events received a clinical diagnosis, and 8.4% of patients with cases of mild/moderate anemia were clinically acknowledged.Testing for iron stores was infrequent; overall, 11.0% of anemia cases received testing for TSAT and 27.4% for ferritin.The proportion of patients who were tested was higher for cases of severe anemia (24.2% received testing for TSAT and 49.4% for ferritin) compared to mild/moderate anemias (10.8% received testing for TSAT and 26.8% for ferritin).Within 6 months of developing anemia, 19.1% of patients received treatment overall, which mostly consisted of iron medication alone (9.9%) and   3).The magnitude of the risk of adverse clinical outcomes was numerically higher after severe anemia events, but also clinically important after mild/moderate anemias.S7) for the risk of MACE and death were 3.22 and 2.00, respectively, which, given the range of HRs observed (Fig. S4), were interpreted as possibly robust to potentially unmeasured confounders.Our conclusions were not modified after excluding early events (Table S8).

Discussion
In this large observational study of patients with CKD stages 3-5 managed in Stockholm's primary health care, approximately one in four (26%) developed anemia before ever visiting a nephrologist.The processes of anemia care were suboptimal, with poor recognition and testing for iron stores.Only 19% of patients with detected anemia received treatment and, although treatment rates increased among those with severe compared with mild/moderate anemia, a large proportion of patients were treated with blood transfusions.Therefore, this study provides evidence of a clinical gap and missed treatment opportunities that warrants modification.Anemia is one of the most common complications in patients with CKD [13], and our study shows that this is also true in those managed in primary care.Overall, 27% of patients with CKD stages 3-5 were excluded from our study due to anemia at study enrollment, and a further 26% developed anemia during follow up.Previous studies in primary care settings have evaluated anemia of CKD and reported similar prevalence proportions [19,26].In a cross-sectional study of primary care units in the UK between 2009 and 2010, 22.2% of patients with CKD stages 3-5 had anemia [19].
In the National Health and Nutrition Examination Survey, a population-representative screening cohort from the US, an analysis of cross-sectional data showed that anemia ranged from 17.4% to 53.4% in patients with CKD stages 3-5 [26].More recent studies in large US health care systems show similar findings [20,27], and although these US data sources did not separate referred from non-nephrologist-referred populations, it is fair to assume that most of these patients would have been identified in connection with primary care consultations.In our study, the average age of patients was 78 years, and while the incidence of anemia reported here aligns with previous studies, data are limited for the very elderly (≥80 years) with CKD, for whom the risk of anemia is elevated [28].In addition to advanced age, and lower eGFR, which intrinsically results in lower erythropoietin production [13], our multivariate risk analysis confirmed the diversity of the risk factors involved in anemia, which also include male sex, and comorbidities such as diabetes mellitus and heart failure [16].Moreover, our results obtained from using a mixed-effect model indicated that there might be some correlation between annual the change in Hb and eGFR, as a rapid and parallel decline was observed for both during the median of 2.5 years.
One key finding in our study was that renal anemia appeared to be under-addressed in primary care.A low proportion (15%) of anemia events in our study were accompanied by a clinical diagnosis, and about 30% of cases of severe anemia lacked a diagnosis.These low rates of recognition may reflect a lack of awareness of the importance of detecting and managing anemia, as supported by other findings of our study.For example, testing of iron stores was seldom performed, even among those patients in whom cases of severe anemia had been detected, which is in clear conflict with guideline recommendations that CKD patients with anemia should be screened and treated for iron deficiency [29,30].Low iron testing rates have also been reported in the United Kingdom and United States [19,20].These gaps in practice are important to note given the low cost of iron tests, and the easy mitigation of iron deficiency through oral or IV iron supplementation, which is a relatively low-risk intervention.
Another key observation of our study is that once anemia was detected and proved to be sustained for at least 6 months, fewer than one in five patients received treatment for anemia.This low rate of treatment provision could be attributed to the perception that cases of mild-to-moderate anemia (16% of which received treatment) may not be as worthy of treatment as cases of severe anemia (treated in 57% of cases).However, we find it disheartening that 43% of cases of severe anemia remained untreated.A contemporary crosssectional analysis from the US showed even lower treatment rates than our study; only 31.7% with severe anemia received treatment [27].
The most common treatment strategies to manage anemia in our analysis were iron replacement therapy (both oral and IV) and blood transfusions.The use of ESAs was infrequent and anecdotal, being given alone or in combination with other strategies to approximately 1% of identified cases of severe anemia.These patterns of care are not dissimilar from those in other countries [19,20], although they highlight a relatively high rate of transfusions as a first line of anemia treatment.This is not aligned with Kidney Disease: Improving Global Outcomes (KDIGO) guidelines, for example, which recommend avoiding or being restrictive with the use of transfusions in renal anemia [29].However, it may represent a deep-rooted practice, as Sweden provides more transfusions than other countries with comparable health care [31].We also observed that at the time of anemia occurrence, the proportion of patients treated with RASi (ACE inhibitors/ARBs) increased by ∼10% from baseline values overall and across different severity categories.In our opinion, the use of RASi and the incidence of anemia are both probably directly related to the severity of CKD (i.e., a lower eGFR), although a plausible biological link between the use of RASi and anemia development has been previously postulated [32].
Our study has direct clinical implications by identifying clinical gaps that require modification.In a recent qualitative study [33], which included interviews with PCPs in the United Kingdom, the authors reported that iron availability and stores data were not always well understood by practitioners.It was also suggested that a lack of clinical expertise and support from specialist care for the safe and cost-effective delivery of ESAs led to concerns over initiating these treatments in primary care.The authors concluded that consequently, a conservative approach to the management of anemia was adopted, and Hb thresholds for the initiation of treatment or referrals to nephrologists were lower than those recommended in guidelines.
We strongly believe that these views also apply in our Swedish setting.Additionally, we speculate that there may be logistical concerns related to prescribing ESAs, including their subcutaneous or IV route of administration, and dose adjustment regimes.Educational programs or decision support systems directed toward PCPs may be beneficial, with a focus on increasing awareness of anemia-associated kidney disease, and informing on effective management strategies.A study in New Zealand [34] implemented a program to identify and treat anemia in patients with CKD who were attending primary care consultations, which consisted of a simple referral and management protocol with remote support by a nephrologist and nurse coordination team.This enabled PCPs to prescribe epoetin beta, and adjust dosages without the need for patients to attend specialist nephrology clinics.However, this was a small study that did not optimize iron status or evaluate safety aspects.It is possible that the new oral formulations of hypoxia-inducible factor-prolyl hydroxylase inhibitors [35,36] may facilitate such programs in the future.
Our analysis showed that patients who developed anemia were at higher subsequent risk of MACE, individual components of MACE (nonfatal stroke, nonfatal myocardial infarction, and nonfatal heart failure) and death.These associations have been reported before [20,27,37] and are likely explained by anemia being a marker for the presence of more severe systemic disease given that a randomized controlled trial did not report improvements in mortality with ESA use compared with placebo [38].However, anemia is still an important marker for the risk of adverse events whose correction has demonstrated marked quality-of-life benefits in numerous trials [38][39][40].Furthermore, the presence of anemia clearly identifies a group of patients at high risk for poor cardiovascular outcomes who may benefit from more focused or improved cardiovascular management.
The strengths of our study include a focus on incident anemia and complete capture of care processes in a single large region with universal health care.This circumvents some limitations that have been identified by previous studies, including their cross-sectional design, and fragmentation of patient cohorts between care or private care practices.We were able to quantify with precision the clinical responses to anemia, and censor participants with proven contact with nephrologists-specialists. The limitations of our study include a reliance on ICD codes for some medical diagnoses, which can provide inaccurate outcomes measures.We represent clinical practice in Stockholm between 2012 and 2018, and although the preceding literature we have identified suggests our findings may be generalizable, caution is advised for extrapolation to other periods, regions, or countries.Moreover, we defined anemia by WHO thresholds, which do not take into account the age-related decline of Hb in adults with advanced age [24,41].Patients may have bought over-the-counter oral iron, and this would not have been captured by our study, but we estimate that the possibility of effective self-medication was low.Finally, in our study, we compared the alignment of clinical responses with guideline recommendations, but we cannot evaluate with certainty whether the responses were adequate.For instance, in multimorbid patients or end-of-life care, clinicians may prioritize the management of conditions other than anemia.Furthermore, we did not identify other relevant responses, such as testing feces-Hb or performing gastroscopy.However, these responses would likely have followed testing of iron stores.
To conclude, this study shows that anemia is a common occurrence among patients with CKD stages 3-5 who are managed in primary care.However, anemia events in this high-risk population were not sufficiently recognized and cared for, as evidenced by a low proportion of formal anemia diagnoses, low rates of screening for iron stores, and low rates of treatment for anemia.Such significant gaps between clinical guidance and current clinical practice likely warrant modification with a holistic approach, given the age and multimorbidity of this population, as well as enhanced communication and cooperation channels between primary and specialist care.

Table 1 .
Baseline characteristics of study participants at baseline and at time of detected anemia occurrence, overall and by severity.
eGFR, estimated glomerular filtration rate; Hb, hemoglobin; IQR, interquartile range; SD, standard deviation.aN= 45,288 at baseline for eGFR measurements.b"Othersignificant disease" includes: dementia, inflammatory bowel disease, liver diseases, and peptic ulcer disease.For definitions of anemia severity categories, please see Methods.Please note that anemia severity categories are not mutually exclusive, and a patient may have developed more than one anemia event of varying severity during follow-up.Fig. 1 Strength of the multivariable association between baseline conditions associated with incident anemia according to the estimated explained relative risk (R 2 , in red), and the proportion of overall explainable log-likelihood (Χ 2 , in blue).Laboratory values were standardized as per standard deviation (SD) increase.ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; CKD, chronic kidney disease; CI, confidence interval; COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate; MRA, mineralocorticoid receptor antagonists; NSAID, nonsteroidal antiinflammatory drug.

Table 2 .
Clinical responses to incident anemia: recognition, testing for iron stores, and treatment patterns.
Abbreviations: CKD, chronic kidney disease; ESA, erythropoietin-stimulating agent; IV, intravenous; TSAT, transferrin saturation.a Within 6 months of the anemia event.For definitions of anemia severity categories, please see Methods.Please note that anemia severity categories are not mutually exclusive, and a patient may have developed more than one anemia event of varying severity during follow-up.and during a median follow-up of 2.6 (IQR 1.1-4.8)years, 11,416 (25.0%) died.After multivariable adjustment, developing anemia was associated with a 1.9-fold (HR 1.91, 95% CI 1.77-2.06)higher risk of MACE and a 1.3-fold (HR 1.33, 95% CI 1.25-1.42)higher risk of death (Table

Table 3 .
Adverse clinical outcomes associated with incident anemia, overall and by anemia severity.Note: MACE was defined as a composite of nonfatal stroke, nonfatal myocardial infarction, nonfatal heart failure, and cardiovascular death.For definitions of anemia severity categories, please see Methods.Please note that anemia severity categories are not mutually exclusive, and a patient may have developed more than one anemia event of varying severity during follow-up.The analysis was adjusted for age, sex, education level, calendar year, diabetes, hypertension, myocardial infarction, heart failure, peripheral vascular disease, cerebrovascular disease, atrial fibrillation, chronic obstructive pulmonary disorder, dementia, inflammatory bowel disease, rheumatoid disease, liver disease, peptic ulcer disease, corticosteroids, immunosuppressants, antibiotics/antivirals/antimycotics, aspirin, non-steroid anti-inflammatory drugs, angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers, mineralocorticoid receptor antagonists, betablockers, oral anticoagulants, and estimated glomerular filtration rate.