15.3 Forecasting Technical and Information Needs
Key Takeaways
- Task A.3 is forecast technical and information needs by linking resources to business needs—not by publishing a technology wish list.
- Forecast capacity (compute, network, licenses, sites), skills (informatics, security, analysts, vendors), and data (identity, grain, lag, new sources) from the written aim.
- A project backlog is expressed demand. A forecast is future need implied by strategy, volume, M&A, and regulation.
- “We need more servers,” “we need an upgrade,” or “we need AI” is not a forecast until it names the business decision, volume, and resource.
- HIMSS does not publish a CPHIMS-official capacity formula. Honesty about assumptions and residual identity beats a precise-looking model.
15.3 Forecasting Technical and Information Needs
Quick Answer: Domain 4 task A.3 is forecast technical and information needs by linking resources to business needs. Start from the written aim. Then name the capacity, skills, and data that aim will consume. “We need more servers” and “we need an EHR upgrade” are shopping lists until they attach to a business need.
You now have organizational aims (A.1) and an environment reading (A.2). Forecasting is the translation step before the IT strategic plan (A.4). If you skip the link to business need, A.4 becomes a vendor roadmap with dates. If you skip environment, you will forecast an employed-system staffing model for an independent medical staff, or an urban digital-front-door for a broadband-poor rural clinic.
Link resources to business needs
A forecast has four clauses. Miss one and you have a wish list.
- Business need quoted from the plan (access, harm, equity, teaching, margin, community benefit).
- What changes in volume, location, decision, or regulation (two clinics, hospital-at-home census, a new value-based contract, a merger).
- Resources consumed or missing (capacity, skills, data, vendors, legal/BAA load, endpoints).
- Horizon and assumption (12-month operational plan versus 24–36-month strategic; what happens if volume is 30% lower).
HIMSS does not publish a CPHIMS-official capacity formula, server-sizing table, or FTE ratio. Do not memorize a fake “CPHIMS forecast = volume × 1.3 licenses.” What the exam tests is the link and the resource classes, with assumptions visible.
| Resource class | Forecast question | Honest link to a business need | Wish-list failure |
|---|---|---|---|
| Capacity — compute, storage, network, cloud, rooms, endpoints, licenses | What volume and latency will the new service create, and where? | Two clinics plus imaging growth imply circuits, PACS cache, scheduling seats, badge/identity at those sites | “Buy more storage” with no volume story |
| Skills — informatics, interface, security, analysts, trainers, vendor managers | Which roles does the aim require that we do not have? | A CDS-heavy harm aim needs clinical informatics and knowledge-maintenance time, not only an analyst to build alerts | “Hire AI people” with no decision the model will support |
| Data — identity, grain, lag, new sources, quality | Which decisions must leaders make, and what can the sources honestly support? | An equity diabetes aim needs EMPI, device-binding, and a published unmatched residual | A new warehouse treated as identity repair |
| Third parties — vendors, BAAs, HIE, payers | Can legal, security, and integration absorb another PHI processor? | A retail-clinic partnership is a BAA, identity, and scheduling problem before it is an app | Signing another NDA and calling it a forecast |
| Sustainment — upgrade testing, backfill, 24×7 coverage | What does year-2 cost after the ribbon cutting? | Hospital-at-home is night coverage and device logistics, not a pilot weekend | Capital-only forecast that ignores run-rate |
Capacity is not only data center. In healthcare it includes exam-room devices, medication-scan reliability, clinic Wi-Fi, license peaks at 8 a.m., and whether the identity system can register uninsured patients at a new site without duplicate MRNs.
Skills are often the binding constraint. A plan that adds two clinics and a virtual-care block without ambulatory analysts, trainers, and a security engineer is not a forecast—it is a future outage. Vendor professional services can cover a spike; they do not replace a sustainment role you never funded.
Data is an A.3 object, not an afterthought for chapter 11. If the business need is a board access KPI, forecast the scheduling-plus-EHR join, the owner, and the match rule. If the need is a value-based contract, forecast claims lag, attribution files, and the residual you will refuse to hide. A lakehouse does not forecast itself into an outcome.
Forecast is not the backlog
The project backlog is demand already expressed—tickets, enhancement requests, replacement of end-of-life hardware. Useful, incomplete. A forecast is future need implied by strategy and environment: the clinics not yet open, the merger not yet closed, the payer contract not yet live, the flu season you already know is coming, the regulation with a dated mandate.
Work both horizons:
- About 12 months: capacity and skills to execute this year’s operational plan. License true-up, interface engineer time, training backfill, device refresh at the new site.
- About 24–36 months: architecture and workforce for multi-year aims. Identity platform, cyber staffing, data products for a risk contract, network to a second county.
Use scenarios, not a single point estimate. Example: the access aim still holds if telehealth stays 10% or jumps to 30%, but the 30% case consumes home-device support, e-visit licensing, and inbox staffing the 10% case does not. Write both. A forecast that cannot survive a volume miss is a budget plea.
Seasonality and campaigns belong in the model. Flu, a community vaccination drive, a new service-line launch, and a go-live that collides with fiscal year-end are forecast inputs. So is M&A: two registration practices and two MRN namespaces are information needs you can see before anyone issues an RFP.
Walk one aim all the way through
Organizational aim: open two neighborhood clinics to cut new-patient wait from 38 days to 21 and improve equity of who gets the slot.
Environment (A.2): mixed employed and affiliated clinicians; sliding-fee patients at one site; thin ambulatory analyst bench.
Linked forecast (A.3), not a shopping list:
- Capacity: scheduling and EHR seats at both sites; network and Wi-Fi; badge and printer identity; referral inboxes sized for the new volume; imaging and lab order routing.
- Skills: ambulatory analyst and trainer time; registration dual-search coaching (duplicate MRN risk); a named security review for a more public site; not “a digital team.”
- Data: wait-time measure from scheduling plus EHR at clinic grain; payer mix and uninsured inclusion so equity is not only insured wait; EMPI rules before any “community dashboard.”
- Assumptions: 16 weeks to hire/train; affiliated physicians may not adopt the enterprise template (A.2); if volume is half the pro forma, license and lease still accrue.
Notice what is absent unless it enables a clause above: a new EHR version, a new data platform, an AI triage bot. Those enter the forecast only when you can name the business decision they change and the resource they consume. An EHR upgrade remains a program candidate for A.4. It is not the forecast.
Scenarios and exam traps
Scenario. Facilities announces two clinics. Infrastructure asks for a storage array “to be safe.” Send it back. Forecast from wait-time and equity aims: seats, circuits, identity at registration, referral capacity, and the analyst who will instrument the board KPI. Storage may appear if imaging volume is real. It is not the need.
Scenario. A director says the forecast is “upgrade the EHR so we are ready for anything.” That is not A.3. Ask which business need the current version cannot support—certification, a closed-loop gap, a new site. If the answer is “readiness,” you have a program looking for an aim.
Scenario. The backlog has 400 enhancements. Leadership calls that the three-year information forecast. It is expressed demand, biased toward whoever files tickets. Build the forecast from aims and volume scenarios, then use the backlog as one input—not as the plan.
Scenario. A value-based contract will start in 14 months. The technical forecast is a new analytics platform. Incomplete. Forecast attribution files, claims lag, EMPI residual, quality-measure specs, and the skills to keep those feeds honest. The platform is optional until those needs are named.
Watch these traps:
- Publishing a technology wish list unlinked to a written business need.
- Treating an EHR upgrade or “AI” as the forecast.
- Equating the enhancement backlog with future need.
- Sizing hardware and ignoring skills, data, and sustainment.
- A single-point volume with no scenario or residual identity.
- Inventing a HIMSS-official CPHIMS sizing formula.
Facilities will open two neighborhood clinics to cut new-patient wait and improve equity. Infrastructure proposes a storage array “to be safe.” What is the correct A.3 response?
Which statement best describes a complete A.3 forecast of technical and information needs?
A director says the three-year technical forecast is “upgrade the EHR so we are ready for anything.” What is missing?