Skip main navigation

Disparity in the Setting of Incident Heart Failure Diagnosis

Originally published Heart Failure. 2021;14:e008538



    Early heart failure (HF) recognition can reduce morbidity, yet HF is often initially diagnosed only after a patient clinically worsens. We sought to identify characteristics that predict diagnosis in the acute care setting versus the outpatient setting.


    We estimated the proportion of incident HF diagnosed in the acute care setting (inpatient hospital or emergency department) versus outpatient setting based on diagnostic codes from a claims database covering commercial insurance and Medicare Advantage between 2003 and 2019. After excluding new-onset HF potentially caused by a concurrent acute cause (eg, acute myocardial infarction), we identified demographic, clinical, and socioeconomic predictors of diagnosis setting. Patients were linked to their primary care clinicians to evaluate diagnosis setting variation across clinicians.


    Of 959 438 patients with new HF, 38% were diagnosed in acute care. Of these, 46% had potential HF symptoms in the prior 6 months. Over time, the relative odds of acute care diagnosis increased by 3.2% annually after adjustment for patient characteristics (95% CI, 3.1%–3.3%). Acute care diagnosis setting was more likely for women compared with men (adjusted odds ratio, 1.11 [95% CI, 1.10–1.12]) and for Black patients compared with White patients (adjusted odds ratio, 1.18 [95% CI, 1.16–1.19]). The proportion of acute care diagnosis varied substantially (interquartile range: 24%–39%) among clinicians after adjusting for patient-level risk factors.


    A large proportion of first HF diagnoses occur in the acute care setting, particularly among women and Black patients, yet many had potential HF symptoms in the months before acute care visits. These results raise concerns that many HF diagnoses are missed in the outpatient setting. Earlier diagnosis could allow for timelier high-value interventions, addressing disparities and reducing the progression of HF.

    What Is New?

    • Documentation of previously undescribed disparities in the setting of incident heart failure diagnosis, including by race and sex.

    • Many patients with acute care diagnoses had potential heart failure symptoms documented at outpatient visits in the months before diagnosis.

    What are the Clinical Implications?

    • Many heart failure diagnoses may be missed in the outpatient setting.

    • Timely diagnosis could permit high-value interventions to start earlier, reducing heart failure progression and addressing the identified disparities.


    More than 1 million Americans are diagnosed with heart failure (HF) annually, with mortality over the first year that exceeds 20%.1 Multiple therapies reduce mortality and improve quality of life among patients with HF, especially among those with reduced ejection fraction. However, curbing the morbidity of HF requires timely diagnosis and initiation of effective therapies.

    International studies have demonstrated incident HF is frequently first diagnosed in the acute care setting (the emergency department or hospital).2–4 They also suggest a concerning trend of increasing frequency of acute care diagnosis over time.3 Incident HF diagnosed in the acute care setting has 1-year mortality exceeding 30%.4 Others have noted the frequency of unrecognized HF in the community.5,6 Earlier recognition and intervention may mitigate HF morbidity. However, the relative frequency of incident HF diagnosis by setting is unclear in the United States. The marked disparities in quality and access of care well-recognized throughout our health care system may also influence initial HF diagnosis patterns.

    In this analysis, we evaluate the frequency of HF diagnosis in the acute care setting versus the outpatient setting in a database of privately insured and Medicare Advantage patients. We identify sociodemographic and clinical characteristics associated with higher odds of acute care diagnosis. Finally, we quantify the variation in acute care diagnosis rates across clinicians and geography.



    Our analysis used the Optum De-identified Clinformatics Data Mart, a database of administrative health claims for members of commercial and Medicare Advantage health plans across all 50 states. This database, which spans from 2003 to 2019, includes ≈17 to 19 million annual patients. This database includes sociodemographic characteristics and medical claims. Data access requests should be sent to Optum; statistical code will be made available upon request.

    Study Population

    We evaluated adult patients 18 years of age or older with an incident HF diagnosis based on the International Classification of Diseases (ICD)-9/ICD-10 diagnosis codes (listed in Table I in the Data Supplement). We defined incident HF based on the absence of previous HF claims during a minimum lookback period during which the patient was continuously insured, excluding patients without a second inpatient or outpatient diagnosis of HF within 1 year of the first diagnosis to reduce the likelihood of including inaccurate diagnoses; researchers working in a variety of health care systems have previously applied similar approaches to defining incident HF, with identical or analogous ICD-9 or ICD-10 codes, lookback periods ranging from 3 to 12 months, and requirements of one or more inpatient or outpatient diagnosis instances for inclusion.7–11 Longer lookback periods reduce the risk of incorrect classification of incident HF but also increase the number of patients excluded from the study. Among all patients with multiple HF diagnoses, the median interval between diagnoses (excluding multiple diagnoses on the same day) was under 180 days for 91% of patients. The maximum interval between any 2 consecutive diagnoses was under 180 days for 54% of patients. We, therefore, used 180 days as the minimum period of database enrollment to identify incident HF. Patients without 180 days before their first diagnosis were excluded. Patients missing all socioeconomic data were also excluded.

    The analysis focused on evaluating HF diagnoses that could potentially be identified before the acute care visit. Accordingly, we excluded patients with concurrent diagnoses of incident HF and acute conditions that can cause sudden-onset HF based on ICD-9/ICD-10 diagnoses codes. This excluded patients with a diagnosis of acute coronary syndrome, myocarditis, aortic dissection, endocarditis, stress-induced cardiomyopathy, and spontaneous coronary artery dissection at the same time as their first HF diagnosis (listed in Table I in the Data Supplement).

    Study Variables

    We extracted demographic, socioeconomic, and medical characteristics from the database. Demographic characteristics included age, sex, and race/ethnicity. The database includes race/ethnicity as a single variable with the following mutually exclusive categories: Asian, Black, Hispanic, and White. Three socioeconomic variables were included in the analysis: educational attainment, economic net worth, and occupation category (crafts/manual labor/trades, homemaker/retired, administrative/clerical/technical/health/civil/military, manager/owner/professional, or unknown). Optum derives socioeconomic data from health information deterministically linked to data licensed from a consumer data vendor.

    We identified medical comorbidities based on ICD-9/ICD-10 diagnoses within 6 months of first HF diagnosis. We also identified the presence of symptoms consistent with HF using medical claim diagnoses. All diagnosis codes are listed in Table I in the Data Supplement. These are nonspecific diagnoses related to abnormalities of breathing, edema, chest pain, palpitations, or syncope. While these symptoms are not specific for HF, their presence before a HF diagnosis suggests the possibility of symptomatic HF.12,13

    The database identifies clinicians or clinician groups who evaluate and treat patients in the inpatient and outpatient settings. We linked patients with new HF to an outpatient primary care clinician, either a physician, nurse practitioner, or physician’s assistant. We restricted to clinicians or clinician groups with the following specialties: internal medicine, family medicine, or general practice. Although the data set includes both identifiers for individual clinicians and clinician groups, we subsequently refer to the attributed group as the clinician alone. In the case of encounters with multiple different clinicians, we attributed the patient to the primary care clinician with the most encounters in the 6 months before their first HF diagnosis. In the context of ties, we used the clinician with the most recent encounter before initial diagnosis.


    The primary outcome was the first diagnosis of HF in the acute care setting: emergency department or inpatient hospitalization. We evaluated variation in the primary outcome based on the state of patient residence and the attributed primary care clinician. We also identified patient-level predictors of acute care diagnosis. As a secondary outcome, we evaluated the proportion of patients with an acute care diagnosis with prior outpatient symptoms suggestive of HF.

    Statistical Analysis

    We described baseline patient characteristics using the mean and SD for continuous variables and percentages for categorical variables. We compared characteristics between those with an initial outpatient versus acute care diagnosis using Cohen d for continuous variables and Cramer V for categorical variables.

    For the primary outcome, we first calculated the proportion of patients with an incident HF diagnosis in the acute care setting. We estimated the association between patient characteristics and acute care diagnosis using a multivariable logistic regression model with adjustment for all characteristics. We also evaluated models with pairwise and 3-way interactions between race/ethnicity, sex, and net worth. We calculated marginal probabilities for each combination of risk factors (eg, women with high net worth). To assess the temporal trend in acute care diagnosis, we used new HF diagnoses between 2004 and 2018. We excluded 2019 given this cohort has <1 year of follow-up for the second HF diagnosis and excluded 2003 as this cohort had limited baseline data. We added quarterly indicators to the model and calculated marginal probabilities for each 3-month period—the probability of initial diagnosis in the acute care setting after adjusting for differences in patient characteristics across time periods.

    We performed several sensitivity analyses. First, we added clinician fixed effects to the model to evaluate whether predictors of acute care diagnosis were similar within clinician practices. Second, we repeated this analysis after limiting the population to patients with a diagnosis of systolic HF to evaluate if predictors were similar among a population more likely to have reduced ejection fraction. Third, we required a longer (1 year) lookback period without a prior HF diagnosis. Finally, to exclude the possibility that observed disparities were driven by differences in access to subspecialty cardiology care, we performed 2 additional sensitivity analyses: first, we adjusted our multivariable regression model for baseline cardiology care (presence and number of clinic visits with a cardiologist between 6 months and 7 days before HF diagnosis); second, we repeated this analysis after excluding all patients with a cardiology visit in this same time frame.

    We quantified variation in the proportion of acute care diagnosis based on patients’ state of residence and their outpatient clinician. To evaluate state-level variation, we added indicators for each individual state. We adjusted for patient-level characteristics to estimate state-to-state variation independent of differences in the patient population. For variation across clinicians, we excluded clinicians with 20 or fewer new HF diagnoses over the study period. We used a mixed-effects model with random intercepts for clinicians. We used random effects for clinicians because the clinicians (and patients) included in this study are a sample of clinicians across the United States. From this model, we predicted the acute care diagnosis proportion for each clinician after adjustment for patient-level characteristics.

    We used Stata 16 (StataCorp LLC, College Station, TX) with a threshold of P≤0.05 for statistical significance. Data access for this project was provided by the Stanford Center for Population Health Sciences Data Core. The Population Health Sciences Data Core is supported by a National Institutes of Health National Center for Advancing Translational Science Clinical and Translational Science Award (UL1 TR001085) and internal Stanford funding. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The study was approved by the Stanford Institutional Review Board.


    We identified 1 745 625 adult patients with incident HF (Figure 1). We excluded 139 475 (8.0%) patients due to missing socioeconomic data. We then excluded patients with a condition known to cause acute-onset HF (eg, acute coronary syndrome) at the same time as their first HF diagnosis and patients without a second HF diagnosis over the subsequent 1 year. There were 959 438 patients with incident HF remaining in the study who formed the analytic cohort. Table II in the Data Supplement compares patient characteristics between the baseline study population and excluded patients.

    Figure 1.

    Figure 1. Diagram of study inclusion and exclusion criteria.bAmong the 261 043 patients with an outpatient primary care visit between 7 d and 6 mo preceding their acute care setting heart failure (HF) diagnosis. SES indicates socioeconomic status.

    Among patients with newly diagnosed HF, 38% were diagnosed in acute care settings. Baseline characteristics for this population overall and by diagnosis setting are shown in the Table. Table III in the Data Supplement displays standardized differences in patient characteristics between those with initial outpatient versus inpatient diagnoses.

    Table. Baseline Patient Characteristics and Association With Acute Care Diagnosis

    CharacteristicOutpatient diagnosis (n=595 514)Acute care diagnosis (n=363 924)Adjusted odds ratio (95% CI) of acute care diagnosis*
    Percentage of total (N) or mean (SD)
    Age, y73.0 (11.9)73.9 (12.0)1.02 per 10 y (1.02–1.03)
    Female sex50.3% (299 674)52.5% (191 033)1.11 (1.10–1.12)
    Enrollment duration before diagnosis, y3.4 (2.9)3.5 (3.0)
    Selected comorbidities
     Alcohol use disorder4.1% (24 126)5.9% (21 472)1.46 (1.43–1.49)
     Asthma14.9% (88 830)17.1% (62 111)1.03 (1.02–1.04)
     Atrial fibrillation30.2% (180 032)39.7% (144 582)1.48 (1.47–1.50)
     Chronic kidney disease25.9% (154 489)32.1% (116 910)1.29 (1.28–1.30)
     Chronic obstructive pulmonary disease30.5% (181 665)37.9% (137 792)1.28 (1.27–1.29)
     Connective tissue disease10.4% (61 810)11.1% (40 266)0.97 (0.96–0.98)
     Dementia5.4% (32 024)4.8% (17 305)0.81 (0.79–0.83)
     Depression23.2% (138 205)25.4% (92 475)1.04 (1.02–1.05)
     Diabetes45.0% (268 133)47.5% (172 733)1.05 (1.04–1.06)
     Hypertension88.3% (525 760)91.2% (332 009)1.16 (1.14–1.18)
     Ischemic heart disease52.9% (314 746)55.5% (202 015)1.02 (1.01–1.03)
     Liver disease10.3% (61 554)11.6% (42 334)1.03 (1.02–1.04)
     Malignancy22.2% (131 974)24.0% (87 247)1.02 (1.01–1.03)
     Other cardiac conduction disorders22.2% (132 128)25.9% (94 390)1.09 (1.08–1.10)
     Other lung disease6.6% (39 047)8.5% (30 945)1.16 (1.14–1.18)
     Peripheral arterial disease30.7% (182 673)32.6% (118 674)0.94 (0.93–0.95)
     Psychotic disorders4.9% (29 293)5.6% (20 455)1.07 (1.05–1.09)
     Supraventricular tachycardia4.4% (25 916)5.3% (19 383)1.01 (0.99–1.03)
     Valvular heart disease35.1% (208 776)39.1% (142 277)1.07 (1.06–1.08)
     Ventricular arrhythmias5.5% (32 727)7.2% (26 144)1.21 (1.19–1.24)
    Social risk factors
      Asian2.5% (14 400)1.9% (6565)0.82 (0.80–0.85)
      Black12.5% (72 498)15.7% (53 652)1.18 (1.16–1.19)
      Hispanic10.8% (62 212)8.5% (29 544)0.79 (0.78–0.81)
      White74.2% (428 718)74.6% (263 943)Reference
      Unknown3.0% (17 686)2.8% (10 220)0.97 (0.95–1.00)
     Net worth
      <$25 00019.1% (113 858)22.8% (82 928)1.39 (1.36–1.41)
      $25 000–$149 00017.9% (106 736)19.0% (69 256)1.26 (1.25–1.28)
      $150 000–$249 00010.8% (64 546)10.9% (39 650)1.20 (1.18–1.22)
      $250 000–$499 00017.2% (102 282)16.2% (59 116)1.13 (1.11–1.14)
      >$500 00020.4% (121 435)17.1% (62 312)Reference
      Unknown14.6% (86 657)13.9% (50 662)1.15 (1.13–1.17)
      Crafts/manual labor/trades2.7% (16 342)2.8% (10 091)1.03 (1.00–1.07)
      Homemaker/retired11.1% (65 909)12.1% (44 048)1.04 (1.01–1.07)
      Administrative/clerical/health/civil/military4.2% (24 772)3.8% (13 656)0.95 (0.92–0.98)
      Manager/owner/professional3.6% (21 699)3.4% (12 262)Reference
      Unknown78.4% (466 792)78.0% (283 867)1.03 (1.01–1.06)
      <12th grade1.0% (826)0.8% (2860)0.88 (0.84–0.92)
      High school33.9% (202 145)36.0% (131 016)1.00 (0.98–1.01)
      <Bachelor degree52.8% (314 349)52.0% (189 392)0.99 (0.97–1.00)
      Bachelor degree11.9% (70 776)10.8% (39 174)Reference
      Unknown0.4% (2418)0.4% (3900)1.04 (0.97–1.11)

    * Odds ratio of acute care diagnosis from multivariable regression model. Model included all listed baseline characteristics except enrollment duration before diagnosis.

    In a multivariable regression model that included medical comorbidities, socioeconomic variables, and demographics, several medical comorbidities were associated with higher odds of acute care diagnosis. For socioeconomic status variables, we found lower net worth was associated with increased odds of acute care diagnosis (adjusted odds ratio [AOR] of net worth of under $25 000: 1.39 [95% CI, 1.36–1.41] compared with over $500 000). For occupation, patients categorized as homemaker or retired in the database were slightly but significantly more likely to be diagnosed in the acute care setting (AOR, 1.04 [95% CI, 1.01–1.07]) than those in the professional category.

    We found evidence of sex and race disparities even after adjustment for clinical comorbidities and markers of socioeconomic status. Women were significantly more likely to be diagnosed in an acute care setting (AOR, 1.11 [95% CI, 1.10–1.12]) compared with men. Black patients were significantly more likely than White patients to be diagnosed in an acute care setting (AOR, 1.18 [95% CI, 1.16–1.19]). This contrasts with Asian and Hispanic patients having lower odds of acute care diagnosis as compared with White patients.

    We repeated the analysis with interactions between race/ethnicity, sex, and net worth (detailed in Tables IVA through IVD in the Data Supplement). With interacted models, we found the proportion of women and men with net worth <$25 000 with acute care setting diagnosis was similar (women 0.42 [95% CI, 0.42–0.42] versus men 0.41[95% CI, 0.41–0.42]) while a higher proportion of women had net worth exceeding $500 000 (women 0.36 [95% CI, 0.36–0.36] versus men: 0.32 [95% CI, 0.32–0.32]; Table IVC in the Data Supplement). A higher proportion of Black patients had acute care setting diagnosis at lower net worth than other race/ethnicity groups (with net worth <$25 000: Black 0.46 [95% CI, 0.46–0.46] versus White 0.42 [95% CI, 0.41–0.42] versus Hispanic 0.35 [95% CI, 0.34–0.36]; Table IVB in the Data Supplement).

    As noted above, we performed multiple sensitivity analyses. First, we repeated the analysis while adjusting for differences across primary care practices to evaluate whether disparities persisted within a given practice. Among the 743 711 patients linked to an outpatient clinician, we found both female sex (AOR, 1.20 [95% CI, 1.19–1.22]) and Black race (AOR, 1.11 [95% CI, 1.09–1.13]) remained associated with acute care diagnosis. Second, we evaluated the 287 254 patients with a systolic HF diagnosis as their first HF diagnosis or within the subsequent year. This cohort had a higher proportion (49.8%) with diagnosis in an acute care setting. Female sex (AOR, 1.18 [95% CI, 1.16–1.20]) and Black race (AOR, 1.24 [95% CI, 1.21–1.26]) remained significantly associated with a higher likelihood of diagnosis in the acute care setting. The results remained consistent for each of the following sensitivity analyses: including patients with a single HF diagnosis, including patients without socioeconomic data, requiring a 1-year lookback period without prior HF diagnosis (Table V and Figures I and II in the Data Supplement), and adjusting for and excluding patients with cardiology visits between 6 months and 7 days before HF diagnosis (Tables VI and VII in the Data Supplement).

    Temporal Trends

    We assessed the annual change in likelihood of acute care diagnosis while adjusting for differences in patient characteristics from 2004 to 2018. The odds of acute care diagnoses increased by 3.2% annually (95% CI, 3.1%–3.3%) after adjusting for differences in patient characteristics across years (Figure 2). Temporal trends were similar for women and Black patients.

    Figure 2.

    Figure 2. Trends in the frequency of first diagnosis of heart failure in the acute care setting. This figure displays the predicted probability of a new heart failure diagnosis being in the acute care setting in each 3-mo period between 2004 and 2018 after adjusting for patient characteristics. The dark blue line is the point estimate, with the gray shading representing the 95% CI.

    Variation by State and Clinician

    Figure 3 depicts significant variation in acute care diagnosis remaining after adding indicators to the model for each individual state. The proportion of acute care diagnosis ranged from 21.2% in California to 46.5% in Vermont after adjusting for patient-level risk factors.

    Figure 3.

    Figure 3. National variation in adjusted percentage of new heart failure diagnosed in the acute care setting.

    We display the overall frequency of primary care clinic visits before HF diagnosis in Table VIII in the Data Supplement. We linked 743 711 patients (77.5%) to primary care clinicians seen in outpatient visits between 6 months and 7 days before their HF diagnosis. Among the 215 727 patients without an outpatient primary care visit over that period, the acute care diagnosis rate was 47.7%. There was a modest inverse correlation between the number of new HF diagnoses per clinician and the adjusted probability of acute care diagnosis (r=−0.15). We found 5677 individual clinicians or clinician groups (subsequently referred to as clinicians) with >20 patients with incident HF. Even after adjustment for patient-level risk factors, the proportion of HF diagnoses made in the acute care setting varied, with an interquartile range of 24.0% to 38.9% (Figure 4).

    Figure 4.

    Figure 4. Variation in the frequency of incident heart failure diagnosis in the acute care setting across primary care clinicians. This figure displays the variation in new diagnosis of heart failure in the acute care setting across primary care clinicians. Each patient was linked to a primary care clinician based on the following hierarchical logic: (1) the clinician who first diagnosed heart failure as an outpatient, (2) the clinician most frequently visited in the 6 months prior to inpatient diagnosis, or (3) the clinician seen soonest before inpatient diagnosis. This analysis was restricted to clinicians with at least 20 new heart failure diagnoses over the study period.

    Prior Symptoms

    Among patients with outpatient primary care visits, we found 46.4% of patients diagnosed in acute care settings had documented symptoms potentially related to HF during outpatient clinic visits between 6 months and 7 days before diagnosis (Table IX in the Data Supplement). Documented symptoms were similar among women (47.2%) and Black patients (46.8%). The most common symptoms were edema (14.5%), cough (12.0%), shortness of breath (11.4%), and chest pain (10.7%).


    In this retrospective analysis of a nationally representative claims database, a large proportion of new HF diagnoses occurred in an acute care setting. This proportion increased over time. We found critical disparities in acute care diagnosis rates; even after adjusting for patient-level risk factors, patients with female sex, Black race, and low net worth had higher acute care HF diagnosis rates. These acute care diagnoses may represent missed opportunities for earlier intervention given nearly half had possible HF symptoms in the 6 months before diagnosis.

    How should one interpret this relatively high—and growing—proportion of new HF diagnoses occurring in the acute care setting? One explanation is that this subset of patients could have a more severe initial disease. However, our results suggest at least some of these acute care diagnoses may have been identified earlier. As noted above, a large proportion had documented symptoms concerning for symptomatic HF in the months before diagnosis. This is consistent with previous studies that found a high prevalence of undetected cardiac dysfunction.5,6 Second, we excluded patients with the causes most likely to cause acute onset of severe HF. Third, the variation in acute care diagnoses across states and clinicians after adjustment for patient-level characteristics suggests that some acute care diagnoses are preventable. In composite, our results suggest acute care diagnosis rates could be reduced if clinicians were more alert to signs and symptoms of HF, particularly among women and Black patients.

    Emerging data suggest it is possible to diagnose HF earlier. A recent analysis by Anderson et al14 compared health care utilization before a first inpatient HF diagnosis to utilization before a first inpatient chronic obstructive pulmonary disease (COPD) diagnosis in Ontario, Canada.14 They found incident HF patients had significantly more health care visits before that hospitalization than a similar COPD population. An analysis of the Diabetes Collaborative Registry by Arnold et al15 of over 1 million outpatients with diabetes found 225 125 patients were on a loop diuretic. However, only 40.9% had a documented diagnosis of HF with variability across clinicians in HF documentation. They also found women were less likely to have documented HF. Even among a population being treated with diuretics, HF recognition is important. Timely HF diagnosis can lead to prescription of other HF-specific therapies, specialty referral, patient counseling, and improved prognostication. These interventions can reduce HF admissions and improve the value of care.

    Analyses of non-US cohorts have also demonstrated a high proportion of acute care diagnoses of incident HF. In an analysis from the United Kingdom, Taylor et al4 found an increase in the proportion of first diagnoses made while inpatient from 28.9% in 2000 to 51.8% in 2010, which subsequently remained stable. In Alberta, Canada, Ezekowitz et al3 found 50.3% of incident HF diagnoses were made in the emergency department or in the hospital.

    We found disparities in HF diagnosis rates that are consistent with previously noted disparities in HF outcomes among low-income and Black patients, even after adjustment for numerous social determinants of health apart from race.16–19 These racial disparities exist in HF care across the disease course: although Black patients are 2 to 3 times more likely than White patients to be diagnosed with HF before the age of 75,1 Black patients with HF are less likely to receive inpatient care for acute HF from a cardiology specialist.20 They are also at higher risk for adverse 30-day clinical outcomes after a HF admission across the socioeconomic spectrum, particularly for individuals living in neighborhoods with greater disadvantage.21 Black patients are also referred for advanced HF therapies later in their disease course.22 Delayed HF recognition and initiation of therapies may reveal another mechanism for differences in morbidity and mortality by race; put differently, this may represent an additional manifestation and pathway of HF-related structural racism.

    In addition to race- and income-based disparities, women were also less likely to be diagnosed with HF before presenting to the acute setting. Prior analyses from the United Kingdom had similar findings.4,18 Our analysis builds on these findings by demonstrating these disparities persist after adjusting for age, comorbidities, and socioeconomic characteristics. This adds to a literature showing worse quality of care for women across multiple process measures: women with HF are less likely to undergo measurement of their left ventricular ejection fraction during a HF admission, to receive a consultation from a cardiologist, or to receive advanced HF therapies in a timely manner.23–25

    The drivers of disparities across race, sex, and income are unclear. We tested whether these differences were driven by the resources available at different clinical practices by adjusting for clinical practices. However, we found differences across race, sex, and economic status persisted within clinical practices. Within a vertically integrated health care system, the Veteran’s Health Administration, Namdaran and Heidenreich found both Black and female veterans were less likely to undergo a transthoracic echocardiogram after having elevated natriuretic peptides suggestive of cardiac dysfunction.26 Our analysis cannot pinpoint whether these inequities are due to limited access to outpatient care, differences in the quality of patient-clinician communication and examination across subgroups, or limited awareness of HF risk and symptoms. It is noteworthy that we analyzed insured patients, raising concerns that inequities would be even larger among the uninsured or marginally insured with less access to care. Further research is needed to better understand the mechanism behind these disparities.

    The increase in frequency of acute care diagnosis over time also merits further investigation. It is unlikely that increasing disease severity explains this trend, with analyses of hospitalized patients suggesting stable or decreased severity.27,28 The composition of the database may have changed over time; however, the increases in inpatient diagnosis rates have also been observed in the United Kingdom.4 These temporal trends may represent increased availability of HF diagnostics (eg, echocardiogram) in the acute care setting, differences in outpatient access or emergency department utilization, or additional yet-to-be-identified explanations.

    There was substantial variation in acute care diagnosis rates across clinicians and states despite controlling for patient-level characteristics. This suggests potential differences in either quality of care or local resource availability. However, there may also be unmeasured patient-level characteristics contributing to this variation. Additionally, differences in patient enrollment in the Optum database across states may contribute to geographic differences. Further research is needed to better understand the root causes of these findings.

    Our study has limitations. First, as with all claims-based analysis, there was limited clinical granularity. We only found 20% of patients in our study had a code of systolic HF, which has been shown to be an imprecise surrogate for HF with reduced ejection fraction.29 The proportion of HF patients with reduced ejection fraction is higher in population studies.30 However, our findings were similar among patients with and without a systolic HF diagnosis. Second, identified clinical comorbidities are potentially incomplete and depend on clinician coding, which may vary across clinicians and patient populations. For this reason, in addition to potential multicollinearity between comorbidities, the association between any individual comorbidity and the odds of acute care diagnosis should be interpreted cautiously. Third, symptoms are only intermittently included on claims when present. Therefore, the absence of a symptom on a claim does not indicate the patient was asymptomatic, suggesting a potentially higher proportion of symptomatic patients before HF diagnosis. Conversely, some nonspecific symptoms, such as chest pain or shortness of breath, may represent a noncardiac problem rather than HF. While there is no a priori reason to suspect that these issues would differ in those diagnosed in the outpatient versus acute care settings, it is not possible to evaluate this with the data currently available. Fourth, this was by definition a population with commercial insurance or Medicare Advantage. Other populations, such as the uninsured or those with Medicaid, may have even higher acute care diagnosis rates. Finally, while we demonstrated substantial variation in acute care diagnosis rates, this analysis was unable to assess the proportion of these diagnoses that could be made earlier or the subsequent impact on morbidity.

    This national study raises concerns that many HF diagnoses are missed in the outpatient setting and exposes important differences in which patients’ diagnoses are missed, highlighting an opportunity to improve the value of care. Making these diagnoses earlier could allow for urgent intervention, preventing the cardiac remodeling and organ dysfunction that accompany delays in treatment. Rapid intervention is critical to prevent morbidity and mortality—both of which remain high in HF, yet can be mitigated with guideline-directed medical therapy. Acute care HF diagnosis portends a worse prognosis than outpatient diagnosis.3 Additional research is needed to evaluate whether earlier recognition and treatment would improve prognosis and also to develop strategies for early screening and diagnosis; the creation of measurement and feedback mechanisms to improve primary care clinicians’ awareness of their own missed diagnoses and subconscious bias is a reasonable starting place. Even if timelier diagnosis does not improve long-term survival, the sociodemographic disparities uncovered deserve to be addressed and further investigated. Patients should know their diagnosis and have the opportunity to prevent disease progression and hospital admission regardless of race, sex, or socioeconomic status.

    Nonstandard Abbreviations and Acronyms


    adjusted odds ratio


    heart failure


    International Classification of Diseases


    The data for this project were accessed using the Stanford Center for Population Health Sciences Data Core. The PHS Data Core is supported by a National Institutes of Health National Center for Advancing Translational Science Clinical and Translational Science Award (UL1 TR001085) and from internal Stanford funding. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

    Supplemental Materials

    Tables I–IX

    Figures I–II

    Disclosures None.


    *A.T. Sandhu and R.L. Tisdale contributed equally.

    The Data Supplement is available at

    For Sources of Funding and Disclosures, see page 867.

    Correspondence to: Alexander Sandhu, MD, MS, Stanford University, 870 Quarry Rd Ext, Palo Alto, CA 94304. Email


    • 1. Virani SS, Alonso A, Benjamin EJ, Bittencourt MS, Callaway CW, Carson AP, Chamberlain AM, Chang AR, Cheng S, Delling FN, et al.; American Heart Association Council on Epidemiology and Prevention Statistics Committee and Stroke Statistics Subcommittee. Heart disease and stroke statistics-2020 update: a report from the American Heart Association.Circulation. 2020; 141:e139–e596. doi: 10.1161/CIR.0000000000000757LinkGoogle Scholar
    • 2. Conrad N, Judge A, Tran J, Mohseni H, Hedgecott D, Crespillo AP, Allison M, Hemingway H, Cleland JG, McMurray JJV, et al.. Temporal trends and patterns in heart failure incidence: a population-based study of 4 million individuals.Lancet. 2018; 391:572–580. doi: 10.1016/S0140-6736(17)32520-5CrossrefMedlineGoogle Scholar
    • 3. Ezekowitz JA, Kaul P, Bakal JA, Quan H, McAlister FA. Trends in heart failure care: has the incident diagnosis of heart failure shifted from the hospital to the emergency department and outpatient clinics?Eur J Heart Fail. 2011; 13:142–147. doi: 10.1093/eurjhf/hfq185CrossrefMedlineGoogle Scholar
    • 4. Taylor CJ, Ordonez-Mena JM, Roalfe AK, Lay-Flurrie S, Jones NR, Marshall T, Hobbs FDR. Trends in survival after a diagnosis of heart failure in the United Kingdom 2000-2017: population based cohort study.BMJ. 2019; 364:l223. doi: 10.1136/bmj.l223CrossrefMedlineGoogle Scholar
    • 5. Mehta H, Armstrong A, Swett K, Shah SJ, Allison MA, Hurwitz B, Bangdiwala S, Dadhania R, Kitzman DW, Arguelles W, et al.. Burden of systolic and diastolic left ventricular dysfunction among hispanics in the United States: insights from the Echocardiographic Study of Latinos.Circ Heart Fail. 2016; 9:e002733. doi: 10.1161/CIRCHEARTFAILURE.115.002733LinkGoogle Scholar
    • 6. Lam CS, Lyass A, Kraigher-Krainer E, Massaro JM, Lee DS, Ho JE, Levy D, Redfield MM, Pieske BM, Benjamin EJ, et al.. Cardiac dysfunction and noncardiac dysfunction as precursors of heart failure with reduced and preserved ejection fraction in the community.Circulation. 2011; 124:24–30. doi: 10.1161/CIRCULATIONAHA.110.979203LinkGoogle Scholar
    • 7. Goyal A, Norton CR, Thomas TN, Davis RL, Butler J, Ashok V, Zhao L, Vaccarino V, Wilson PW. Predictors of incident heart failure in a large insured population: a one million person-year follow-up study.Circ Heart Fail. 2010; 3:698–705. doi: 10.1161/CIRCHEARTFAILURE.110.938175LinkGoogle Scholar
    • 8. Desai RJ, Mahesri M, Chin K, Levin R, Lahoz R, Studer R, Vaduganathan M, Patorno E. Epidemiologic characterization of heart failure with reduced or preserved ejection fraction populations identified using Medicare claims.Am J Med. 2021; 134:e241–e251. doi: 10.1016/j.amjmed.2020.09.038CrossrefMedlineGoogle Scholar
    • 9. Akwo EA, Kabagambe EK, Wang TJ, Harrell FE, Blot WJ, Mumma M, Gupta DK, Lipworth L. Heart failure incidence and mortality in the Southern Community Cohort Study.Circ Heart Fail. 2017; 10:e003553. doi: 10.1161/CIRCHEARTFAILURE.116.003553LinkGoogle Scholar
    • 10. Prasada S, Rivera A, Nishtala A, Pawlowski AE, Sinha A, Bundy JD, Chadha SA, Ahmad FS, Khan SS, Achenbach C, et al.. Differential associations of chronic inflammatory diseases with incident heart failure.JACC Heart Fail. 2020; 8:489–498. doi: 10.1016/j.jchf.2019.11.013CrossrefMedlineGoogle Scholar
    • 11. Ahmad FS, Chan C, Rosenman MB, Post WS, Fort DG, Greenland P, Liu KJ, Kho AN, Allen NB. Validity of cardiovascular data from electronic sources: the Multi-Ethnic Study of Atherosclerosis and HealthLNK.Circulation. 2017; 136:1207–1216. doi: 10.1161/CIRCULATIONAHA.117.027436LinkGoogle Scholar
    • 12. Maddox TM, Januzzi JL, Allen LA, Breathett K, Butler J, Davis LL, Fonarow GC, Ibrahim NE, Lindenfeld J, Masoudi FA, et al.. 2021 update to the 2017 ACC expert consensus decision pathway for optimization of heart failure treatment: answers to 10 pivotal issues about heart failure with reduced ejection fraction: a report of the American College of cardiology solution set oversight committee.J Am Coll Cardiol. 2021; 77:772–810. doi: 10.1016/j.jacc.2020.11.022CrossrefMedlineGoogle Scholar
    • 13. Mann DL, Chakinala M. Heart failure: pathophysiology and diagnosis.Jameson J, Fauci AS, Kasper DL, Hauser SL, Longo DL, Loscalzo J. eds. In: Harrison’s Principles of Internal Medicine, 20e. McGraw-Hill.Google Scholar
    • 14. Anderson K, Ross HJ, Austin PC, Fang J, Lee DS. Health care use before first heart failure hospitalization: identifying opportunities to pre-emptively diagnose impending decompensation.JACC Heart Fail. 2020; 8:1024–1034. doi: 10.1016/j.jchf.2020.07.008CrossrefMedlineGoogle Scholar
    • 15. Arnold SV, Jones PG, Beasley M, Cordova J, Goyal A, Fonarow GC, Seman L. Heart failure documentation in outpatients with diabetes and volume overload: an observational cohort study from the diabetes collaborative registry.Cardiovasc Diabetol. 2020; 19:212. doi: 10.1186/s12933-020-01190-6CrossrefMedlineGoogle Scholar
    • 16. Nayak A, Hicks AJ, Morris AA. Understanding the complexity of heart failure risk and treatment in black patients.Circ Heart Fail. 2020; 13:e007264. doi: 10.1161/CIRCHEARTFAILURE.120.007264LinkGoogle Scholar
    • 17. Husaini BA, Mensah GA, Sawyer D, Cain VA, Samad Z, Hull PC, Levine RS, Sampson UK. Race, sex, and age differences in heart failure-related hospitalizations in a Southern state: implications for prevention.Circ Heart Fail. 2011; 4:161–169. doi: 10.1161/CIRCHEARTFAILURE.110.958306LinkGoogle Scholar
    • 18. Lawson CA, Zaccardi F, Squire I, Ling S, Davies MJ, Lam CSP, Mamas MA, Khunti K, Kadam UT. 20-year trends in cause-specific heart failure outcomes by sex, socioeconomic status, and place of diagnosis: a population-based study.Lancet Publ Health. 2019; 4:e406–e420. doi: 10.1016/S2468-2667(19)30108-2CrossrefMedlineGoogle Scholar
    • 19. Glynn P, Lloyd-Jones DM, Feinstein MJ, Carnethon M, Khan SS. Disparities in cardiovascular mortality related to heart failure in the United States.J Am Coll Cardiol. 2019; 73:2354–2355. doi: 10.1016/j.jacc.2019.02.042CrossrefMedlineGoogle Scholar
    • 20. Breathett K, Liu WG, Allen LA, Daugherty SL, Blair IV, Jones J, Grunwald GK, Moss M, Kiser TH, Burnham E, et al.. African Americans are less likely to receive care by a cardiologist during an intensive care unit admission for heart failure.JACC Heart Fail. 2018; 6:413–420. doi: 10.1016/j.jchf.2018.02.015CrossrefMedlineGoogle Scholar
    • 21. Patel SA, Krasnow M, Long K, Shirey T, Dickert N, Morris AA. Excess 30-day heart failure readmissions and mortality in black patients increases with neighborhood deprivation.Circ Heart Fail. 2020; 13:e007947. doi: 10.1161/CIRCHEARTFAILURE.120.007947LinkGoogle Scholar
    • 22. Breathett K, Allen LA, Helmkamp L, Colborn K, Daugherty SL, Blair IV, Jones J, Khazanie P, Mazimba S, McEwen M, et al.. Temporal trends in contemporary use of ventricular assist devices by race and ethnicity.Circ Heart Fail. 2018; 11:e005008. doi: 10.1161/CIRCHEARTFAILURE.118.005008LinkGoogle Scholar
    • 23. Klein L, Grau-Sepulveda MV, Bonow RO, Hernandez AF, Williams MV, Bhatt DL, Fonarow GC. Quality of care and outcomes in women hospitalized for heart failure.Circ Heart Fail. 2011; 4:589–598. doi: 10.1161/CIRCHEARTFAILURE.110.960484LinkGoogle Scholar
    • 24. Ehrmann Feldman D, Xiao Y, Bernatsky S, Haggerty J, Leffondré K, Tousignant P, Roy Y, Abrahamowicz M. Consultation with cardiologists for persons with new-onset chronic heart failure: a population-based study.Can J Cardiol. 2009; 25:690–694. doi: 10.1016/s0828-282x(09)70528-8CrossrefMedlineGoogle Scholar
    • 25. Morris AA, Cole RT, Laskar SR, Kalogeropoulos A, Vega JD, Smith A, Butler J. Improved outcomes for women on the heart transplant wait list in the modern era.J Card Fail. 2015; 21:555–560. doi: 10.1016/j.cardfail.2015.03.009CrossrefMedlineGoogle Scholar
    • 26. Namdaran P, Heidenreich PA. Use of echocardiography following an elevated B-type natriuretic peptide level.Circ: Cardiovasc Qual Outcomes. 2017; 10:A093.LinkGoogle Scholar
    • 27. Hamo CE, Fonarow GC, Greene SJ, Vaduganathan M, Yancy CW, Heidenreich P, Lu D, Matsouaka RA, DeVore AD, Butler J. Temporal trends in risk profiles among patients hospitalized for heart failure.Am Heart J. 2021; 232:154–163. doi: 10.1016/j.ahj.2020.11.015CrossrefMedlineGoogle Scholar
    • 28. Parizo JT, Kohsaka S, Sandhu AT, Patel J, Heidenreich PA. Trends in readmission and mortality rates following heart failure hospitalization in the veterans affairs health care system from 2007 to 2017.JAMA Cardiol. 2020; 5:1042–1047. doi: 10.1001/jamacardio.2020.2028CrossrefMedlineGoogle Scholar
    • 29. Heidenreich PA, Natarajan S, Bahrami H. Accuracy of administrative coding to identify reduced and preserved left ventricular ejection fraction.J Card Fail. 2019; 25:486–489. doi: 10.1016/j.cardfail.2019.01.019CrossrefMedlineGoogle Scholar
    • 30. Lam CS, Donal E, Kraigher-Krainer E, Vasan RS. Epidemiology and clinical course of heart failure with preserved ejection fraction.Eur J Heart Fail. 2011; 13:18–28. doi: 10.1093/eurjhf/hfq121CrossrefMedlineGoogle Scholar