Why EHRs Still Miss 40% of Chronic Condition Complexity

And what that means for care, coding, and AI

Electronic health records (EHRs)—often still called EMRs—were supposed to give us a complete, computable picture of each patient. 

In reality, when you look closely at problem lists, diagnosis codes, and multimorbidity measures, a tough pattern emerges: 

The typical EHR view misses a huge chunk of chronic-condition complexity — easily 30–50%, depending on how you measure it. 

Studies show that: 

• Problem list completeness for common chronic diseases varies widely, with only ~60–80% of known conditions actually present in the problem list across institutions. 

• For some high-impact diseases like stage 3–4 chronic kidney disease, only 16% of patients with lab-defined CKD had CKD documented on the problem list. 

• Reviews of multimorbidity measurement note that EHR systems underestimate the burden of multiple chronic conditions because of missing or out-of-date data. 

• Validation work on ICD-10 administrative data finds under-coding of co-existing conditions, especially asymptomatic ones, with sensitivities as low as ~40% for some comorbidities. 

Put together, it’s reasonable for a health system to assume that on the order of ~40% of chronic-condition “signal” is invisible if you rely only on problem lists and billing codes in the EHR. 

This article unpacks why that happens, what “40%” practically means, and what providers should demand from their data and AI tools going forward. 

1. EHRs capture encounters; patients live in trajectories

Most EHR data models were built around encounters: 

• Each visit gets diagnoses, meds, orders. 

• Problem lists accumulate items over time. 

• Claims data records what was billed. 

But chronic disease complexity lives in trajectories: 

• Kidney function drifting down over five years 

• HF symptoms waxing and waning with changing meds 

• COPD, obesity, depression, and diabetes interacting in a feedback loop 

Multimorbidity research using large EHR datasets repeatedly finds that: 

• Chronic conditions cluster in complex ways, and 

• These patterns are hard to see if you only look at one code list at a time. 

EHRs absolutely store the raw ingredients of these trajectories—labs, vitals, meds, notes—but the structured “front window” (problem list + last set of visit diagnoses) is often a blurry, partial summary. 

That’s the first way EHRs miss complexity: they weren’t designed to reason over whole patient timelines. 

2. Problem lists: better than nothing, far from complete

Problem lists are supposed to be the single source of truth for a patient’s major diagnoses. 

The reality is more complicated. 

2.1 What the data say about problem list completeness 

Across multiple studies: 

• A multi-site analysis of EHRs at 10 health systems found problem list completeness for chronic diseases ranged from 60.2% to 99.4%, with an average around 78%. 

• In outpatient settings, common chronic diseases showed both missing diagnoses and duplications on problem lists, with completeness and quality varying by condition. 

• For chronic kidney disease, only 16% of patients with lab-defined stage 3–4 CKD had CKD recorded on the problem list; multiple studies have described CKD as under-recognized and under-documented in EHRs. 

A 2022 study of patients with heart failure in the emergency department found that the EHR problem list captured multimorbidity with moderate-to-good accuracy, but still misclassified an average of two comorbidity domains per patient when compared with a reference list based on full records. 

In other words: problem lists are directionally useful, but they systematically lose detail and drop conditions, especially: 

• “Background” conditions that feel less urgent (e.g., CKD, anemia) 

• Historical conditions that no one is sure whether to keep 

• Issues owned by other specialties 

  
  

2.2 Why problem lists fall behind 

Qualitative work on problem list use points to familiar issues: 

• Unclear ownership (“Who maintains this? Primary care? Specialist? Hospitalist?”) 

• Time pressure during visits 

• Fear of cluttering the list, making it unusable 

• Lack of feedback or incentive tied directly to maintaining a high-quality list 

When no one “owns” the problem list as a shared clinical asset, it decays. Chronic-condition complexity ends up scattered across: 

• Old notes 

• Scanned hospital records 

• Lab trends 

• Imaging reports 

…while the official snapshot looks deceptively simple. 

3. Codes and claims: the second blindness 

If problem lists understate complexity, maybe claims data makes up for it? 

Not quite. 

3.1 Under-coding of comorbidities in administrative data 

Recent evaluations of ICD-10 administrative data compared with chart review show: 

• Sensitivity for individual comorbidities ranging from ~40–90%, with some conditions consistently under-coded. 

• Asymptomatic or lower-priority conditions (e.g., some chronic diseases that weren’t the main reason for admission) are particularly likely to be missed. 

• Under-coding of co-existing conditions has persisted or worsened over time in some settings, even as coding systems and financial incentives evolved. 

  
  

A 2017 analysis of co-existing conditions in administrative data explicitly highlights under-coding as a major issue, especially for conditions that don’t drive immediate billing or are overshadowed by the primary diagnosis. 

3.2 Why claims data underestimates multimorbidity 

Claims are optimized for payment, not clinical nuance: 

• Only a handful of diagnoses are coded per encounter. 

• Coders are under time pressure and prioritize principal diagnoses and high-impact comorbidities. 

• Historical context (e.g., “HF with preserved EF vs reduced EF,” CKD staging, diabetic nephropathy vs “diabetes + CKD”) may never be fully expressed in codes. 

  
  

Multimorbidity researchers emphasize that large datasets based only on codes systematically underestimate the true number and complexity of chronic conditions in patient populations. 

So if you stack: 

• Imperfect problem lists, and 

• Under-coded claims 

  
  

…you can easily end up blind to 30–50% of the “long tail” of chronic disease complexity in your panels. 

4. What “40% of complexity” actually looks like

“40%” is not a universal constant. It’s a pragmatic shorthand for what multiple lines of evidence are telling us: 

• Problem list completeness across systems averaging ~78% (so ~22% of conditions missing), with much worse documentation for certain diseases (e.g., CKD). 

• Multimorbidity reviews and methods papers warning that EHR-based counts underestimate disease burden because of missing or outdated entries. 

• Administrative data validation studies finding substantial under-coding of co-existing conditions, especially those that are asymptomatic or less central to the index encounter. 

• Disease-specific work (e.g., CKD, HF) showing dramatic gaps between objective signal in labs/imaging and diagnoses formally recorded in the EHR. 

  
  

If you’re a health system that: 

• Relies mainly on problem lists and diagnosis codes to understand your population, and 

• Hasn’t invested in longitudinal, phenotype-based analytics across labs, meds, and notes 

• 
  

…it is entirely realistic that 30–50% of your patients’ chronic-condition interactions, severities, and etiologies are effectively invisible to your dashboards and AI tools. 

Concretely, that 40% might include: 

• CKD that is evident from eGFR trends but never labeled 

• Diabetic neuropathy, retinopathy, or nephropathy documented narratively but not coded as such 

• HFpEF vs HFrEF distinctions buried in echo reports 

• Depression, frailty, or cognitive impairment that shows up in free text and functional assessments but not in structured codes 

• 
  

The patient is complex; your EHR summary says “HTN, DM2, OA.

5. Why EHRs miss complexity: five structural reasons 

5.1 Fragmented data within the EHR 

A 2023 study of chronic condition identification in Swiss primary care EHRs showed that no single EHR component (problem list, diagnoses, prescriptions, or notes) was sufficient to capture prevalence; accurate estimation required integrating multiple components. 

In real systems, chronic disease evidence is split across: 

• Problem lists 

• Encounter diagnoses 

• Labs and imaging 

• Orders and medications 

• Free-text notes and scanned documents 

Most analytics pipelines still pull from one or two of these, not all. 

5.2 Asymptomatic and “background” conditions 

Conditions like early CKD, anemia, or mild heart failure often don’t drive the encounter agenda. Research repeatedly shows: 

• CKD is under-recognized and under-documented despite objective lab evidence. 

• Co-existing conditions without acute symptoms are especially likely to be missing from administrative data. 

If it’s not the “chief complaint” or the main driver of billing, it often doesn’t show up in the structured summary. 

5.3 Time and incentives 

Problem list and code quality improve when organizations: 

• Provide gap reports and feedback to clinicians 

• Link problem lists to billing and quality incentives 

• Make problem-oriented charting an integral part of workflow 

  
  

But in many settings, clinicians receive: 

• Little direct feedback on problem list completeness, and 

• Stronger signals about throughput and relative value units (RVUs) 

  
  

So complexity remains in the chart, not in the fields your analytics see. 

5.4 Legacy data and migration issues 

Longitudinal studies of chronic disease in EHRs note that: 

• Conditions diagnosed before a certain date (e.g., before HIE go-live) may not be present in the EHR dataset at all. 

• Migrations and system changes can strand older diagnoses in “historic” or unstructured formats. 

So your EHR may only truly reflect the last few years of complexity, even for 20-year disease trajectories. 

5.5 The combinatorial explosion of multimorbidity 

As multimorbidity network studies point out, once patients have 5–10 chronic conditions, the number of possible combinations explodes. Many combinations appear in only a handful of patients. 

Most EHR interfaces and registries simply aren’t designed to represent these complex patterns in ways that are easy to see or query, so signals get flattened. 

6. Why this matters: care, contracts, and AI 

6.1 Clinical care and population health 

If your EHR misses 40% of chronic condition complexity, you risk: 

• Under-recognizing high-risk patients who actually have dense multimorbidity. 

• Missing opportunities for disease management (e.g., CKD or HF clinics) because patients don’t meet coded criteria. 

• Designing population health programs on an incomplete denominator, which dilutes impact. 

• 
  

Trials of EHR-based population health management for CKD emphasize the need to identify high-risk patients using labs and care gaps, not just diagnoses on problem lists. 

6.2 Risk adjustment and value-based contracts 

Under-captured complexity has direct financial implications: 

• Risk scores that don’t reflect actual multimorbidity can lead to under-payment for truly complex patients. 

• Conversely, naive efforts to “code harder” on a shaky data foundation risk over-coding or mis-coding, which auditors are increasingly scrutinizing. 

  
  

If you’re negotiating or operating under risk-bearing contracts while your EHR only sees 60% of the true chronic disease landscape, you are making strategic decisions with partial information. 

6.3 AI and clinical decision support 

Most first-generation AI tools (including some LLM-based solutions) are trained or evaluated on the structured fields that underrepresent complexity: 

• Problem lists 

• Claim codes 

• Simple comorbidity counts 

  
  

Reviews of multimorbidity and EHR-based AI warn that these limitations can lead to biased models that underestimate risk, especially for patients whose complexity lives in under-documented or fragmented data. 

If we don’t fix the data layer, “AI for chronic disease management” just learns a sharper version of the same blind spots. 

7. What providers should demand from their data (and AI)

 

Closing the 40% gap is not about telling clinicians to “code more.” It’s about building infrastructure that can see reality. 

At minimum, providers should expect that any serious population health, risk, or AI solution: 

1. Builds patient-level phenotypes, not just counts of codes 

   1. CKD defined by eGFR trends + albuminuria + diagnoses, not codes alone. 

2. Integrates multiple EHR components 

   1. Problem lists + encounter diagnoses + labs + meds + notes, as demonstrated in chronic disease prevalence work using EHR components. 

3. Audits itself against chart review or enriched reference standards 

   1. Periodically comparing phenotype outputs with clinician-reviewed samples, as ICD-10 validation and multimorbidity studies do. 

4. Surfaces “silent” complexity 

   1. Explicitly highlights patients whose labs, meds, or notes suggest additional chronic conditions or greater severity than codes reflect. 

5. Keeps longitudinal context front and center 

   1. Designs interfaces and AI explanations around trajectories and timelines, not just static snapshots. 

6. Respects clinical workflow and ownership 

   1. Aligns with how clinicians think about the problem list—as a shared, longitudinal clinical resource—and gives them tools (not nagging) to keep it meaningful. 

      
      

The goal isn’t a perfect EHR; it’s an accurate enough longitudinal model of each patient to support good care, fair payment, and trustworthy AI.

References  

1. Wright A, et al. Problem list completeness in electronic health records: A multi-site study and assessment of success factors. J Am Med Inform Assoc. 2015. 

2. Wang ECH, et al. Characterizing outpatient problem list completeness and duplications in common chronic diseases. Appl Clin Inform. 2020. 

3. Samal L, et al. Electronic problem list documentation of chronic kidney disease in a primary care setting. BMC Nephrol. 2014. 

4. Jolly SE, et al. CKD in an electronic health record problem list: Quality of care and outcomes. Clin J Am Soc Nephrol. 2014. 

5. Chase HS, et al. Under-documentation of chronic kidney disease in the electronic health record. J Am Med Inform Assoc. 2010. 

6. King BL, et al. Accuracy of the electronic health record’s problem list in identifying multimorbidity in patients with heart failure. PLoS One. 2022. 

7. Pan J, et al. Assessing the validity of ICD-10 administrative data in coding comorbidities. (Alberta, Canada cohorts). 2025. 

8. Peng M, et al. Impact of co-existing conditions, death status and number of observation years on the incidence and undercoding of chronic conditions. Health Informatics J. 2017. 

9. Suls J, et al. Measuring Multimorbidity: Selecting the Right Instruments. Ann Behav Med. 2021. 

10. Meier R, et al. Importance of different electronic medical record components in identifying chronic conditions in primary care. Swiss Med Wkly. 2023. 

11. Chua YP, et al. Definitions and Prevalence of Multimorbidity in Large Database Studies. Int J Environ Res Public Health. 2021. 

12. Beridze G, et al. Patterns of multimorbidity in primary care electronic health record data: A systematic review. Digit Health. 2024. 

13. Monchka BA, et al. The effect of disease co-occurrence measurement on multimorbidity networks. BMC Med Res Methodol. 2022. 

14. Zhou X, et al. Cooperative documentation: the patient problem list as a multipurpose tool. AMIA Proc. 2012. 

15. Jhamb M, et al. Electronic health record–based population health management intervention to reduce CKD progression. JAMA Intern Med. 2024. 

16. Sperl-Hillen JA, et al. A Cluster Randomized Clinical Trial in Primary Care Clinics: Closing care gaps in CKD. 2024