Digital accessibility for elderly patients is a design problem
Digital accessibility for elderly patients in hospitals is usually framed as a training problem — patients need help learning the portal. It is more often a design problem, because the portal was built and tested by people in their thirties on recent phones with good eyesight, stable connectivity and daily app fluency. The patient who cannot complete registration is not failing; the interface is.
This matters more in healthcare than almost anywhere else because the age distribution is inverted. The heaviest users of hospital services are precisely the group least served by digital-first design, and a hospital that moves registration, payment or reports online without accommodating them has shifted work onto its most vulnerable patients.
The test is simple and rarely run: can a seventy-five-year-old with reading glasses, an older phone and no help complete the task. Until that has been observed rather than assumed, accessibility claims are aspirational.
Type, contrast and touch targets
Start with the physical layer. Body text below roughly sixteen pixels is difficult for many older readers, and low-contrast grey-on-white — popular because it looks refined — is the single most common accessibility failure in healthcare interfaces. Aim for strong contrast on text that carries instructions, and let the visual restraint live in layout rather than in faded type.
Touch targets need to be generous. Reduced fine motor control, tremor and dry skin all make small targets unreliable, and a mis-tap in a payment flow is not a minor irritation. Adequate spacing between adjacent actions matters as much as the size of each, particularly where one of them is destructive.
Respect system settings. A patient who has already enlarged text on their phone has told you what they need; an interface that overrides it with fixed sizing has ignored the most reliable accessibility signal available. Test the portal at increased system text size and check that nothing becomes unreachable.
Baseline checks before launching a patient-facing screen
- Readable body text with strong contrast on instructions
- Touch targets large enough and spaced apart
- Layout survives the patient's enlarged system text size
- No information conveyed by colour alone
- Every field labelled visibly, not only by placeholder text
- Works on an older mid-range phone and a slow connection
OTP fatigue and the authentication trap
One-time passwords are the default authentication method in Indian digital services and they are a significant barrier for elderly patients. The sequence — leave the app, find the message, read six digits, remember them, return, enter them before expiry — involves reading small text under time pressure and short-term memory load. Each additional OTP in a journey multiplies the failure rate.
Count the OTPs in your flows. It is common to find three or four in a single journey: one to log in, one to link a record, one to authorise a payment, one to view a report. Consolidate ruthlessly. A longer session for low-risk actions, with re-authentication only for genuinely sensitive operations, is both more usable and no less secure in practice than one that trains users to expect constant prompts.
Where OTPs are unavoidable, make the mechanics forgiving: generous expiry, an obvious resend, autofill support so the code does not have to be transcribed, and an error message that says what to do rather than that something went wrong. Also verify the number reaches the patient rather than an adult child, because otherwise the authentication is silently delegated.
Assisted digital: the staffed counter that never goes away
Every digital patient service needs an assisted path, permanently. Not as a transition arrangement during rollout, but as a standing channel, because a meaningful share of patients will never complete these tasks independently and their care must not degrade for it. Hospitals that plan to retire the counter once adoption reaches a target are planning to exclude someone.
The assisted path should reach the same system, not a parallel one. A staff member completing the form on the patient's behalf at a counter produces the same record, the same identifiers and the same downstream benefits; a paper fallback that is keyed in later produces reconciliation work and a different data quality. Design the staff-assisted mode as a first-class mode of the same application — a platform such as HealUDoc can expose the same registration flow at a counter as it does on the patient's phone, so the assisted route is not a second-class copy that drifts out of step.
Record that assistance was given and by whom. This matters for accountability where the staff member entered clinical or financial information on the patient's behalf, and it gives you the honest denominator: what proportion of digital transactions were actually completed by patients themselves.
Caregiver proxy access and its consent implications
In practice, adult children and spouses manage a great deal of elderly patients' health administration, and hospitals mostly accommodate this informally — sharing information with whoever accompanies the patient, or letting a family member use the patient's login. Informality is convenient until it is not, because it removes the patient's ability to control what is shared and leaves no record of who saw what.
A proper proxy model gives the caregiver their own credentials, linked to the patient with the patient's recorded authorisation, and scoped. A caregiver may need appointments, bills and general instructions without needing the full clinical record, and the patient should be able to set that boundary and change it later. Under the DPDP Act framework, consent must be informed and withdrawable, which a shared password cannot deliver — nobody can withdraw access from someone who knows their password.
Handle the difficult cases explicitly: patients who lack capacity, situations where the caregiver's interest may conflict with the patient's, and elder abuse. Delegated access should be revocable by the patient without the caregiver being notified, and staff should have a route to raise concerns. A system such as HealUDoc can carry scoped delegated access with an audit trail, but the policy questions about who may be granted what remain the hospital's to answer.
“Half our elderly patients were logging in as themselves from a son's phone in another city. Giving the son his own limited access was safer for everyone, including the son.”
Kiosks in the actual lobby, not the design studio
Self-service kiosks fail for environmental reasons more often than interface reasons. Glare from a glass frontage makes the screen unreadable at the height an older patient stands. There is nowhere to put a bag or a walking stick. There is no seat, and the interaction takes four minutes. The queue behind creates time pressure that guarantees errors.
Place and physically design kiosks for the people who will use them: reachable from a seated position, out of direct glare, with a surface for belongings, and with room for an attendant to stand alongside without blocking the path. A kiosk that requires a patient to stand unsupported for several minutes has excluded a large group before the software loads.
Keep a staff member within sight of the kiosk bank during busy periods. The most common reason kiosks go unused is not fear of technology but the absence of anyone to ask, and a single roving assistant converts more usage than any interface improvement.
Test with the patients you actually serve
Usability testing with staff is not usability testing. Staff know the vocabulary, the workflow and what the abbreviations mean, and their fluency conceals every problem the design has. Recruit eight or ten real patients across the age and literacy range you serve, give them a task, and watch without helping.
Watching without helping is the hard part and the whole point. The instinct to guide someone who is struggling destroys the observation you came for. Note where they hesitate, what they read aloud, what they tap first, and where they give up — the give-up point is the most valuable data in the session.
Repeat it whenever the flow changes materially. Accessibility is not a certification that is earned once; it degrades every time a new field, a new consent screen or a new verification step is added by a team optimising for something else.
