DCHS Insights CX Human-Centered Design @ DCHS PROTOTYPE no real data
DCHS Insights CX How We Screen

Universal screening: three approaches

Screening and referral are one design problem. We need a light, consistent way to surface needs beyond the program someone came for—and a credible path from every yes to information, consent, a referral, and follow-through. These prototypes make the tradeoffs concrete enough to test.

The planning question

What is the lightest screening and referral flow that can surface important needs without costing clients unnecessary time, privacy, agency, or trust?

What this carries forward

At the 2025 leadership offsite, the team proposed the idea of a ONE DCHS — Standard Inclusive Intake process, which emerged from the Aboul & Hossina persona journey. This section carries that idea forward by separating it into decisions that can be designed and piloted: who answers, what everyone is asked, when deeper review is appropriate, who controls the next step, and how the referral loop closes.

Why now: HCS is planned to replace Harmony later this year, creating a stronger case-management foundation for cross-program referrals. Infrastructure alone will not surface needs or coordinate care; DCHS still needs a light, scalable approach to screening, consent, routing, and follow-through.

Use the prototypes to design the flow

The goal is not to pick the most polished screen. It is to decide how needs are surfaced and what happens next.

  1. Take stock of existing screening processes

    Map where and how DCHS screens today, identify pain points for clients and staff, ask how these processes could be strengthened, and capture what current practice can teach the broader design.

  2. Walk each persona through every approach

    Mark where someone would stop, decline, misunderstand, disclose too much, or still have a need missed.

  3. Compare the bets

    Focus on scale, depth, staff involvement, and client control rather than the visual design of a particular screen.

  4. Follow every yes

    Specify the information, consent, referral, acknowledgment, follow-through, and response when service capacity is unavailable.

  5. Choose a small pilot

    Name program-to-program connections and teams that are specific and ready enough to test first.

Three screening strategies

The literature converges on three distinct approaches, and every screening tool trades administrative efficiency against depth of insight. A layered combination—self-service, a short standard set, and deeper staff-led review—is a strong working hypothesis. Use the comparison to test when each layer is appropriate and where none is.

1

Interactive Self-Service

Clients screen themselves, anonymously, and get a list of programs they may qualify for.

What it is

A public-facing web tool with branching logic: pick the kinds of help you're looking for, answer a short sequence, get a customized program list. Later questions depend on earlier answers, so nobody answers items that don't apply.

What it is good at

Scale. It moves initial triage from staff to the client, works anonymously, manages expectations before a formal application — and it is the only approach that can meaningfully reduce application abandonment, by carrying answers forward into the application itself.

Where it falls short

It depends on digital literacy, internet access, and accurate self-reported income. It determines programmatic eligibility, not need — and it only reaches people who already came looking.

Scalability High
Standardization Moderate
Depth of insight Low
2

Small Set of Standard Questions

A short, fixed set of validated questions asked of everyone, at every intake point.

What it is

Identical wording, identical response options, every time, regardless of which program the person is entering.

What it is good at

Standardization. Because everyone answers the same questions the same way, the results support population-level analysis, geographic hot-spotting, and automated referral triggers — and the wording maps cleanly to LOINC and ICD-10 Z-codes for health-system interoperability.

Where it falls short

Diagnostically shallow by design. A positive screen says a problem exists, not what is causing it — food insecurity could be money, transportation, or a food desert. It is triage, and must be paired with human navigation downstream.

Scalability Moderate
Standardization Highest
Depth of insight Moderate
3

Holistic Review

A caseworker-administered assessment scoring a family across many life domains, repeated over time.

What it is

A structured interview on an ordinal scale, re-administered at intervals to track movement — the tier that surfaces causal relationships a checklist misses.

What it is good at

Depth and outcome measurement. It can show that a childcare gap is driving employment instability, which is driving housing insecurity — and because the same instrument repeats, it produces defensible longitudinal evidence of whether a program actually moved anyone.

Where it falls short

Administrative intensity: a trained interviewer, real time per family, severe reporting bias if self-administered. It scales poorly as a front door and belongs with clients already in supportive programs. And many matrices in circulation were never validated — choose one with published reliability and budget for rater training.

Scalability Low
Standardization Moderate
Depth of insight Maximum

Decisions the prototypes need to support

These questions turn a comparison of screens into a screening-to-referral design.

What is the smallest question set DCHS would genuinely ask of everyone, every time?

Which questions are too sensitive, too burdensome, or only useful when a service is actually available?

What moves someone from the short screen to a deeper review?

Does the client, staff member, or system control routing—and how can a client decline without penalty?

Where should the workflow live so it becomes part of practice rather than another platform to check?

What closes the referral loop, and who is accountable when capacity is unavailable?

Which program-to-program connections are ready enough to pilot first?

Prototype candidates

One concrete example per approach, chosen to prototype against — with the shortlist of alternatives kept brief.

Primary example · Interactive Self-Service

ACCESS NYC

Org Website ↗

30+ City, State, and Federal programs, regardless of immigration status, completable anonymously in roughly ten minutes.

The part worth studying is architectural, not visual: the screener is a single-page app calling a rules API, the rules live in a public repository, and the same rules are exposed through a Benefits Screening API for outside developers. NYC's own retrospective is candid that the technology for reuse is solid — the policy and incentives for other agencies to adopt it never materialized.

We built our own take on this — the DCHS Prototype (1b)

How Can We Help is our jurisdiction's counterpart to ACCESS NYC, with a deliberately different posture: minimal information collection, speed, self-determination over which categories to walk through, the ability to exit at any time with something useful in hand, and a generated guidance document to keep. It avoids complex income data collection entirely — the produced guide shows the eligibility requirements, leaving that review to the client.

Open the DCHS Prototype →
Also looked at
  • MyFriendBen — The one with momentum — open source, rules via PolicyEngine, returns dollar values; six states, NC stood theirs up locally with 211. No Virginia instance yet.
  • Bridge to Benefits — long-running nonprofit predecessor (CDF-Minnesota); dated design.
The Virginia contrast: CommonHelp “Am I Eligible?”

Worth walking through as a contrast, not a model — it asks for a great deal of income and expense detail up front.

Walk through it ↗
Primary example · Small Set of Standard Questions

CMS AHC HRSN Screening Tool

The instrument (PDF) ↗

Ten core questions across five domains — housing instability, food insecurity, transportation, utilities, interpersonal safety — plus sixteen supplemental questions across eight more. 7th-grade reading level, free, and widely adopted enough that our data would be comparable to other jurisdictions.

Two things to know: it is an assembly of separately-owned items (CMS published the per-question citation table — keep it in our documentation), and the federal reporting scaffolding that once justified it has largely been withdrawn — the business case has to be local and operational.

Known gaps: no childcare domain (none of the comparable instruments have one — a federal-survey item is included in our companion item bank), and it omits income, legal needs, digital access, and justice involvement; behavioral health coverage is thin.

We built our own take on this — the AHC-HRSN screener (03)

All thirty items as a working screener: the ten core questions asked of everyone, sixteen supplemental items offered as optional areas rather than required ones, and three local additions. Every question is transcribed verbatim and carries its own source citation behind an info icon, so the licensing requirement travels with the instrument instead of crowding the screen.

Open the AHC-HRSN screener →
Also looked at
Primary example · Holistic Review

Colorado Family Support Assessment 2.0

Already in use at DCHS Fact sheet (PDF) ↗

Our Office of Community Services administers the CFSA 2.0 today — so the open question is not whether to adopt it, but how a Tier 3 assessment connects to whatever we build upstream.

Three parts: Part A scores 14 domains (eight form an Economic Self-Sufficiency scale, two a Health scale); Part B is the FRIENDS Protective Factors Survey; Part C captures the family's own readiness for change. Administered by staff trained in Motivational Interviewing, with follow-ups at 31–90 day intervals.

An OMNI Institute study found strong psychometrics (~96% coding accuracy, published in Research on Social Work Practice). One fit test from FRCA's own checklist: it assumes strength-based case management follows the baseline — without that follow-up, it is the wrong tool for measuring outcomes.

We built our own take on this — the CFSA 2.0 field companion (04)

Part A as an advocate's companion rather than a form: all fourteen domains, every scoring anchor from 5 down to 1, and the Prevention Line drawn where the instrument draws it. The definitions stay open beside the criteria because they are scoring rules, and nothing invents a level name or a composite score the source does not define.

Open the CFSA 2.0 companion →
Also looked at
What to carry forward

A draft screening-to-referral design

The useful result is not a winning prototype. Carry forward the choices needed to build a coherent flow and a small pilot that can test it.

Reference library

There is no shortage of instruments — the CMS panel behind the AHC tool started from 200+ questions across 50+ instruments, in 2016. This review picks one concrete example per approach to prototype against, not a survey of the field.

SIREN screening tool comparison ↗

Side-by-side view of the most widely used tools, with domain coverage and question wording.

AJPM systematic review (Henrikson et al., 2019) ↗

Systematic review of social risk screening instruments.

Kaiser Permanente Washington SDH tools review ↗

Searchable database of instruments with psychometric detail.

DCHS Insights CX

This planning site is for internal use. Please enter the access password to continue.