What Might You Want in a Complete Longitudinal Medical Record?

We in health information technology anxiously await the arrival of complete semantic interoperability so we can help patients assemble a complete longitudinal medical record for themselves. When we achieve this state of health information exchange, we will still need to address higher-order challenges including: 1) the validity of diagnoses entered into a medical record, 2) assessing the relevance of information that is gathered intermittently, if at all and 3) determining what textual elements within clinical documentation map to a specific disease or its severity. Even if we believe these additional barriers can be overcome, what questions might the complete longitudinal medical record answer that we can’t answer today?

Some would argue that we have a greater interest in our complete longitudinal financial record. When I reviewed at a few personal finance-oriented websites, they recommend tracking

  • Identity and property ownership – proof of citizenship (e.g., passport), birth certificate, marriage certificate, adoption papers, Social Security card, titles, deeds
  • Current finances – credit cards, bills, brokerage accounts, current mortgage and other loans
  • Finances when you cannot speak for yourself – durable power of attorney, insurance (disability and long-term care insurance)
  • Information to protect you when facing an unexpected legal event – tax records, earlier mortgages, property records
  • Information for your loved ones after you die – will, trust, safe deposit box keys, life insurance documents, burial instructions, obituary information

If I applied these categories to healthcare, I might collect the following*:

  • Identity – where I have received care in the past over what date range including medical record number(s) and patient portal login information
  • Current health – active problem list, medication allergies, medication list with preferred pharmacy information, primary care provider and other members of the care team (perhaps clinicians I have seen in the last 12 months), measurements or behaviors I am tracking to manage my current health (e.g., weight, blood pressures)
  • Managing my future health when I cannot speak for myselfdurable power of attorney for healthcare, living will with end-of-life preferences, health insurance, preferences for diagnostic uncertainty, treatment side effects and quality-of-life
  • Information to protect me when facing an unexpected medical event – operative reports, hospital discharge summaries, consults for medical conditions with long-term implications, immunizations
  • Information for my loved ones after I die – preferences about donating organs or my entire body for clinical care or research

*This approach assumes there is nothing in my medical record that I (or my family members) might be embarrassed or otherwise penalized for sharing with others (e.g., whole genome sequence). I might use a separate process to manage those data.

We may want a longitudinal medical record to help inform others about our healthcare journey at its effect on our overall life. Symptom management, independence (activities of daily living and instrumental activities of daily living), quality-of-life, quantity of life, sensory input, cognitive function and financial burden all seem important. Disease-specific metrics may provide additional information to others interested in specific challenges with a particular disease or constellation of diseases. The International Consortium for Health Outcomes Measurement has outlined 28 standard sets of measurements for specific health conditions (e.g., one survey for macular degeneration [Brief Impact of Vision Impairment Questionnaire (Brief IVI)], four surveys for heart failure [Kansas City Cardiomyopathy Questionnaire-Short Version, New York Heart Association Functional Classification, PROMIS Physical Function Short Form 4a – Patient-Reported Outcome Measurement Information System, Patient Health Questionnaire-2], five surveys for “older” persons [SF-36, ASCOT toolkit, UCLA 3-item loneliness scale, Barthel index, Canadian Study on Health & Aging Clinical Frailty Scale]). All of these metrics, including our own preferences, may change over time, suggesting additional value in tracking these measurements.

Asking individuals to complete the same survey over time introduces new challenges. First, as our understanding of health changes over time, surveys may be expanded, but information collected about new points of interest are subject to each individual’s recollection of prior events. Second, not everyone who receives a survey goes on to complete it. Published survey responses have ranged from 70-ish% (2018 Nurses’ Health Study publication about metabolic unhealthy phenotypes and cardiovascular disease risk, 2018 Cancer Prevention Study-3 [disclaimer: I am enrolled in this study]) to >99% (Physicians’ Health Study I, 11 years of annual 19-question surveys). People who do not complete surveys at regular intervals may be systematically different from those who do, limiting a study’s generalizability. Third, people who do complete surveys may make mistakes or have someone else complete the survey, reducing their validity.

In additional to our understanding of what patient-specific information to track among patients with a single condition, two other points are worth noting. One, patients with multimorbidity may be able to provide additional information to others about how they prioritize specific behaviors or treatments. Wei et al. found dementia, lung disease and arthritis to have a larger effect on physical functioning than hypertension, glaucoma, hip replacement or myocardial infarction (without heart failure) using self-reporting of chronic conditions using a multimorbidity-weighted index. Two, tracking caregiver well-being (both perceived strain and psychological distress) may also provide meaningful information to other patients and their care teams.

Our push for a longitudinal healthcare record may lead to frustrations like we experienced after publishing the human genome. For an individual concerned about their own healthcare, tracking relevant behaviors and metrics with treatment preferences and some measure of aversion to unwanted outcomes may help healthcare members place medical records in the appropriate context to determine the next best step in diagnosis, treatment or long-term management. Patients interested in helping others through decision-making may consider sharing their longitudinal information through trusted sources.