The Amplification Gap

The measurement problem

Operational staffing decisions are only as good as the demand data behind them. In service lines that respond to consults — spiritual care, social work, case management, and others — that data is almost always a count of initial encounters: the page, the call, the new consult. It is easy to capture and easy to trend, and it feels like a complete picture of demand.

However, an initial encounter is just the tip of the iceberg. Much of the actual labor these services perform is continuation work: the follow-up visits, the family that needs to be called back repeatedly, the case that stays open for days. When that continuation work is logged separately — or, as is common, not logged at all — leaders end up sizing their workforce to meet the visible half of the demand. The denominator is wrong. So is the staffing that follows from it.

This is not a new observation. In nursing, where workload measurement is comparatively mature, researchers have documented that patient-classification and volume-based systems systematically underestimate true demand by omitting the "hidden" activities that resist easy counting.¹ The problem is simply more acute in consult-responding services, where the tracking is thinner and the hidden work is a larger share of the whole.

The consequence is subtle, but expensive. Two sources can generate the same number of initial calls while producing very different amounts of total work, and a call-based view cannot tell them apart. Staffing decisions made to meet the loud, countable demand signal quietly under-resource the sources and units where the real workload accumulates.

Case study: pastoral care at a Level I trauma center

Spiritual care is a revealing place to look, in part because the field still lacks broad consensus on how to staff to demand. Even among top-ranked hospitals, benchmarking finds chaplain staffing varies nearly twofold across comparable institutions: roughly 2.5 department full-time equivalents (FTEs) per 100 average daily census, but with a standard deviation of 1.3. The spread points to staffing decisions set by history and budget more than by any shared measure of need.²

To test how large the gap between consult and follow-up volume can be, I built a structured, de-identified dataset of pastoral-care activity over a 12-week period in Q4 2025.³ Each row represented one patient encounter with a chaplain, coded by referral source and by phase: Initial (the event that opened chaplain involvement) or Follow-Up (subsequent continuation work tied to an earlier episode).

The headline numbers looked ordinary. Of 152 events, acute clinical events (traumas, codes, rapid responses) drove most of the calls — about 56% of encounters — and the Emergency Department was the busiest location by a wide margin. A leader glancing at call volume would reasonably conclude that acute events and the ED are where the workforce should be concentrated. That is, in fact, how the service was staffed.

The phase data told a different story.

1. Follow-up work was substantial and structurally invisible. Continuation events made up 28% of all activity (43 of 152 encounters) — but because the department's shift reports captured only initial calls, none of that work appeared in the dashboards the department actually used to reason about demand. Roughly one in four events were simply off the books.

2. The rate of downstream work varied more than fourfold by source. To measure this, I calculated the amplification ratio: follow-ups generated per initial event by referral source⁴. By this metric, acute clinical events were high-volume but largely one-and-done, generating 0.21 follow-ups per initial. Clinical-staff referrals from nurses and physicians were far lower in call volume, but each initial spun up nearly a full follow-up: 0.91 per initial. That is a 4.2× difference (rate ratio 4.24; 95% CI 2.17–8.29; p < 0.001).⁵ Put differently, clinical-staff referrals were only 28% of initial calls but generated 47% of all follow-up work.⁶

3. The pattern reordered the service's real workload. The signal a call count alone would have missed showed up cleanly at the unit level. The ICU produced only 11% of initial calls over the quarter, but its amplification ratio was 0.92 follow-ups per initial, nearly the highest of any unit. Staff to call volume, and the ICU looks like a rounding error; staff to total work, and it is a priority. The ED remained the single largest source of work overall,⁷ but call volume alone systematically under-ranked the smaller, higher-amplification sources and units.

4. The hidden load was predictable. Continuation work did not behave like the emergencies that triggered it. Only 16% of follow-ups occurred overnight (midnight–6am); the majority fell in daytime and early-evening hours, with afternoon being the heaviest block.⁸ Unlike acute calls, follow-up work is schedulable. That distinction matters for what kind of staffing coverage it needs.

Why this generalizes

This study demonstrates the amplification gap for one service line at one hospital. What it suggests is broader. The mechanism behind the gap — a low-volume source generating disproportionate downstream demand that call-based tracking cannot see — is not specific to chaplaincy. It is a property of how consult-responding services are measured, and any service that (a) responds to consults and (b) logs initial encounters more reliably than continuation work is structurally exposed to the same blind spot. Social work and clinical case management are the obvious next candidates: they share the consult-and-follow-up structure, and much of their labor is sustained engagement with existing patients rather than new referrals.

The framework: measure amplification, then staff to it

Closing the amplification gap does not require new software or a research study. It requires four incremental changes at the service level.

1. Log the second event. Widen what counts as a recordable event so that continuation work is captured alongside initial calls, each tagged with a phase flag (Initial vs. Follow-Up) and the source that generated it. This is the single change that makes the hidden half of the workload visible.
2. Capture rough duration. Counts understate continuation work because follow-ups vary widely in length. Even a coarse time estimate per event (for example, ≤15 min / 15–60 / 60+) converts a raw event count into a record of time spent — much more useful for staffing decisions.
3. Compute the amplification ratio. Measure follow-ups per initial referral, by source and by unit. Add this ratio to your dashboards and track the trends over 7, 30, and 90 days. Monitoring amplification surfaces exactly the sources and units that punch above their call-volume weight, and it does so with a number that service leadership can compare, trend, and defend.
4. Match coverage to the shape of the load, then set a threshold and re-measure. Demand has a shape, not just a size. Acute, emergent work is unpredictable and needs 24/7 on-call readiness; continuation work is higher-total but schedulable and daytime-weighted. Those two profiles call for different coverage instruments, deployed in proportion to the measured load, with a defined threshold (continuation hours per week, or the ratio crossing a set line) at which coverage steps up or down. Recompute each period so staffing tracks the metric rather than a one-time estimate.

From metric to staffing decision

In the case above, the framework points to a concrete and defensible set of moves. Service line leadership should reallocate proactive rounding and coverage toward the high-amplification pockets the ratio exposes — clinical-staff consults and the ICU — rather than toward raw call-count leaders alone. And because the continuation load is real, sizable, and schedulable, the service should add a second dedicated responder for rounding and follow-up, separate from the 24/7 on-call that covers emergent events.

The data even suggests the shift. Continuation work is not a nine-to-five load: it builds through the day, peaks in the afternoon, and stays elevated into the evening, with more than half of all follow-ups (56%) falling between noon and midnight and only 16% overnight. A strict business-hours responder would miss the evening entirely — nearly a quarter of the continuation load. A responder scheduled across the afternoon and evening instead, from approximately 12–9pm, would absorb the bulk of the demand.

Limitations

This analysis is descriptive and exploratory, so its findings are deliberately bounded.⁹ It reflects a single service line at a single site over a single quarter (n = 152 events), so its ratios are specific to this setting and should not be read as generalizable magnitudes. Duration was not captured in the source data, so the workload figures are event counts rather than hours — almost certainly an understatement of continuation cost. Documentation practices varied across shifts and staff. And because a follow-up was attributed to the source recorded on that follow-up event, the amplification ratio reflects association rather than a traced causal chain from a specific initiating encounter.

However, these caveats bear on how far the finding travels, not on whether it holds here. Within this dataset, the amplification effect is large and statistically unambiguous. The findings strongly support running similar studies in other service lines and clinical settings.

Conclusion

You cannot staff for work you do not measure. Call-based tracking guarantees that a consult-responding service line measures only half of its job. No single metric can reliably capture the whole of that workload; some of it will always resist counting.¹ But that is no reason to keep missing the part that counts cleanly. Tracking the amplification ratio allows service leadership to read the shape of the continuation workload and size coverage to it. The alternative is to keep staffing to the loudest signal while the real work accumulates, unseen, in the follow-ups nobody counted.