REFERENCE / HEOR TABLE SHELLS
HEOR table shells
Nine publication-ready empty tables for a standard retrospective HEOR database study — from cohort attrition through adjusted cost, HCRU, and survival analyses. Pre-specify the format before you see the data.
A table shell is an empty table: every row label, every column header, and every footnote written out, with the cells left blank. Analysts fill in the numbers only after the statistical analysis plan (SAP) is finalized and locked.
Pre-specifying shells in the SAP matters for one reason: it forces every analytic decision — which comparisons, which subgroups, which summary statistics — to happen before anyone has seen how the numbers turn out. That separation is what keeps a retrospective HEOR study from drifting into results-driven table building, and it is usually the first thing a reviewer or regulator checks.
The nine shells below cover the standard sequence for a comparative retrospective claims study: attrition, baseline balance, unadjusted HCRU and cost, adjusted cost models, and a survival analysis. Cells use placeholder notation (XX.X, XXX (XX.X), X,XXX) so the shape and formatting of every table is visible without real data.
Shell 01 / 09
Table 1. Cohort selection and attrition
Purpose: document the sequential application of each inclusion and exclusion criterion, so a reader can see exactly how the source population narrows to the final analytic cohort.
| Selection criterion | N | % of previous step | % of initial |
|---|---|---|---|
| Source population (all patients in database, study period) | X,XXX,XXX | — | 100.0 |
| ≥1 qualifying diagnosis or index drug claima | XXX,XXX | XX.X | XX.X |
| Age ≥18 years at index date | XXX,XXX | XX.X | XX.X |
| 12-month continuous medical and pharmacy enrollment pre-indexb | XX,XXX | XX.X | XX.X |
| No pre-index exclusion diagnosis | XX,XXX | XX.X | XX.X |
| ≥6-month continuous enrollment post-index or deathb | XX,XXX | XX.X | XX.X |
| Assigned to Cohort A or Cohort B per index exposure | XX,XXX | XX.X | XX.X |
| Final analytic cohort | XX,XXX | XX.X | XX.X |
a Index date is defined as the date of the first qualifying diagnosis or index drug claim during the identification period.
b Continuous enrollment allows gaps of ≤30 days without breaking the enrollment span.
Shell 02 / 09
Table 2. Baseline demographic characteristics
Purpose: summarize demographic composition of the overall cohort and each comparison group, and flag imbalance between groups using standardized mean difference (SMD).
| Characteristic | Overall (N=X,XXX) | Cohort A (N=X,XXX) | Cohort B (N=X,XXX) | SMDa |
|---|---|---|---|---|
| Age, mean (SD), years | XX.X (XX.X) | XX.X (XX.X) | XX.X (XX.X) | X.XXX |
| Age, median (IQR), years | XX.X (XX.X–XX.X) | XX.X (XX.X–XX.X) | XX.X (XX.X–XX.X) | — |
| Age band, n (%) | ||||
| 18–34 | XXX (X.X) | XXX (X.X) | XXX (X.X) | X.XXX |
| 35–44 | XXX (X.X) | XXX (X.X) | XXX (X.X) | X.XXX |
| 45–54 | XXX (X.X) | XXX (X.X) | XXX (X.X) | X.XXX |
| 55–64 | XXX (X.X) | XXX (X.X) | XXX (X.X) | X.XXX |
| 65–74 | XXX (X.X) | XXX (X.X) | XXX (X.X) | X.XXX |
| 75+ | XXX (X.X) | XXX (X.X) | XXX (X.X) | X.XXX |
| Female, n (%) | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Race / ethnicity, n (%) | ||||
| White | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Black | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Hispanic | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Asian | XXX (X.X) | XXX (X.X) | XXX (X.X) | X.XXX |
| Other / unknownb | XXX (X.X) | XXX (X.X) | XXX (X.X) | X.XXX |
| US region, n (%) | ||||
| Northeast | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Midwest | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| South | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| West | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Unknown | XX (X.X) | XX (X.X) | XX (X.X) | X.XXX |
| Payer type, n (%) | ||||
| Commercial | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Medicare Advantage | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Medicaid | XXX (X.X) | XXX (X.X) | XXX (X.X) | X.XXX |
| Other | XX (X.X) | XX (X.X) | XX (X.X) | X.XXX |
| Index year, n (%) | ||||
| 2021 | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| 2022 | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| 2023 | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| 2024 | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| 2025 | XX (X.X) | XX (X.X) | XX (X.X) | X.XXX |
a SMD >0.10 is interpreted as a meaningful imbalance between Cohort A and Cohort B.
b All characteristics measured over the 12-month baseline window ending on the index date.
Cells with n<11 are masked per data use agreement.
Abbreviations: SD, standard deviation; IQR, interquartile range; SMD, standardized mean difference.
Shell 03 / 09
Table 3. Baseline clinical characteristics
Purpose: summarize comorbidity burden and pre-index treatment and cost profile, to assess clinical comparability of the two cohorts before any outcome analysis.
| Characteristic | Overall (N=X,XXX) | Cohort A (N=X,XXX) | Cohort B (N=X,XXX) | SMD |
|---|---|---|---|---|
| Charlson Comorbidity Index (Quan), mean (SD)a | X.X (X.X) | X.X (X.X) | X.X (X.X) | X.XXX |
| CCI category, n (%) | ||||
| 0 | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| 1 | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| 2 | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| 3+ | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Elixhauser comorbidity count, mean (SD) | X.X (X.X) | X.X (X.X) | X.X (X.X) | X.XXX |
| Individual comorbidities, n (%) | ||||
| Hypertension | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Type 2 diabetes | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Chronic kidney disease | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| COPD | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Heart failure | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Obesity | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Depression / anxiety | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Hyperlipidemia | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Baseline medication classes, n (%) | ||||
| ACE inhibitors / ARBs | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Statins | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Anticoagulants | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Opioids | XXX (XX.X) | XXX (XX.X) | XXX (XX.X) | X.XXX |
| Baseline all-cause healthcare costs, mean (SD), USD | $X,XXX (X,XXX) | $X,XXX (X,XXX) | $X,XXX (X,XXX) | X.XXX |
a Comorbidity indices computed from 12-month baseline claims using the Quan ICD-10-CM adaptation of the Charlson and Elixhauser algorithms.
Abbreviations: CCI, Charlson Comorbidity Index; COPD, chronic obstructive pulmonary disease; ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; SD, standard deviation; SMD, standardized mean difference.
Shell 04 / 09
Table 4. All-cause healthcare resource utilization during follow-up
Purpose: compare unadjusted, all-cause utilization across care settings between cohorts, normalized for variable follow-up time.
| HCRU category | Cohort A (N=X,XXX) | Cohort B (N=X,XXX) | P valueb |
|---|---|---|---|
| Inpatient admissions | |||
| Patients with ≥1, n (%) | XXX (XX.X) | XXX (XX.X) | 0.XXX |
| Admissions per patient, mean (SD) | X.XX (X.XX) | X.XX (X.XX) | 0.XXX |
| PPPMa, mean (SD) | X.XXX (X.XXX) | X.XXX (X.XXX) | 0.XXX |
| Length of stay, days, mean (SD) | X.X (X.X) | X.X (X.X) | 0.XXX |
| Emergency department visits | |||
| Patients with ≥1, n (%) | XXX (XX.X) | XXX (XX.X) | 0.XXX |
| Visits per patient, mean (SD) | X.XX (X.XX) | X.XX (X.XX) | 0.XXX |
| PPPM, mean (SD) | X.XXX (X.XXX) | X.XXX (X.XXX) | 0.XXX |
| Physician office visits | |||
| Patients with ≥1, n (%) | XXX (XX.X) | XXX (XX.X) | 0.XXX |
| Visits per patient, mean (SD) | XX.X (X.X) | XX.X (X.X) | 0.XXX |
| PPPM, mean (SD) | X.XX (X.XX) | X.XX (X.XX) | 0.XXX |
| Other outpatient services | |||
| Patients with ≥1, n (%) | XXX (XX.X) | XXX (XX.X) | 0.XXX |
| Encounters per patient, mean (SD) | X.XX (X.XX) | X.XX (X.XX) | 0.XXX |
| PPPM, mean (SD) | X.XX (X.XX) | X.XX (X.XX) | 0.XXX |
| Pharmacy fills | |||
| Patients with ≥1, n (%) | XXX (XX.X) | XXX (XX.X) | 0.XXX |
| Fills per patient, mean (SD) | X.X (X.X) | X.X (X.X) | 0.XXX |
| PPPM, mean (SD) | X.XX (X.XX) | X.XX (X.XX) | 0.XXX |
a PPPM = per patient per month, computed as count ÷ (days of follow-up ÷ 30.44), to normalize for variable follow-up time between patients.
b Count comparisons use the statistical test appropriate to the outcome's distribution (e.g., Wilcoxon rank-sum for continuous counts, chi-square or Fisher's exact for proportions).
Abbreviations: HCRU, healthcare resource utilization; PPPM, per patient per month; SD, standard deviation.
Shell 05 / 09
Table 5. Disease-specific healthcare resource utilization during follow-up
Purpose: isolate utilization directly attributable to the disease of interest, using the same category structure as the all-cause HCRU table for direct comparison.
| Disease-specific HCRU category | Cohort A (N=X,XXX) | Cohort B (N=X,XXX) | P value |
|---|---|---|---|
| Disease-specific inpatient admissions | |||
| Patients with ≥1, n (%) | XXX (XX.X) | XXX (XX.X) | 0.XXX |
| Admissions per patient, mean (SD) | X.XX (X.XX) | X.XX (X.XX) | 0.XXX |
| PPPM, mean (SD) | X.XXX (X.XXX) | X.XXX (X.XXX) | 0.XXX |
| Length of stay, days, mean (SD) | X.X (X.X) | X.X (X.X) | 0.XXX |
| Disease-specific ED visits | |||
| Patients with ≥1, n (%) | XXX (XX.X) | XXX (XX.X) | 0.XXX |
| Visits per patient, mean (SD) | X.XX (X.XX) | X.XX (X.XX) | 0.XXX |
| PPPM, mean (SD) | X.XXX (X.XXX) | X.XXX (X.XXX) | 0.XXX |
| Disease-specific physician office visits | |||
| Patients with ≥1, n (%) | XXX (XX.X) | XXX (XX.X) | 0.XXX |
| Visits per patient, mean (SD) | X.XX (X.XX) | X.XX (X.XX) | 0.XXX |
| PPPM, mean (SD) | X.XX (X.XX) | X.XX (X.XX) | 0.XXX |
| Disease-specific other outpatient services | |||
| Patients with ≥1, n (%) | XXX (XX.X) | XXX (XX.X) | 0.XXX |
| Encounters per patient, mean (SD) | X.XX (X.XX) | X.XX (X.XX) | 0.XXX |
| PPPM, mean (SD) | X.XX (X.XX) | X.XX (X.XX) | 0.XXX |
| Disease-specific pharmacy fills | |||
| Patients with ≥1, n (%) | XXX (XX.X) | XXX (XX.X) | 0.XXX |
| Fills per patient, mean (SD) | X.X (X.X) | X.X (X.X) | 0.XXX |
| PPPM, mean (SD) | X.XX (X.XX) | X.XX (X.XX) | 0.XXX |
a Disease-specific is defined as any claim carrying a qualifying diagnosis in the primary diagnosis position; a sensitivity analysis repeats the table using a qualifying diagnosis in any position.
b Disease-specific pharmacy fills are limited to the pre-specified drug list for the condition under study.
Abbreviations: HCRU, healthcare resource utilization; ED, emergency department; PPPM, per patient per month; SD, standard deviation.
Shell 06 / 09
Table 6. All-cause healthcare costs during follow-up
Purpose: compare unadjusted, all-cause cost burden across medical and pharmacy categories between cohorts, in both raw and per-patient-per-month terms.
| Cost category, USDa | Cohort A (N=X,XXX) | Cohort B (N=X,XXX) | P value |
|---|---|---|---|
| Total costs | |||
| Mean (SD) | $XX,XXX (XX,XXX) | $XX,XXX (XX,XXX) | 0.XXX |
| Median (IQR) | $X,XXX (X,XXX–XX,XXX) | $X,XXX (X,XXX–XX,XXX) | 0.XXX |
| PPPMb, mean (SD) | $X,XXX (X,XXX) | $X,XXX (X,XXX) | 0.XXX |
| Total medical costs | |||
| Mean (SD) | $XX,XXX (XX,XXX) | $XX,XXX (XX,XXX) | 0.XXX |
| PPPM, mean (SD) | $X,XXX (X,XXX) | $X,XXX (X,XXX) | 0.XXX |
| Inpatient | $XX,XXX (XX,XXX) | $XX,XXX (XX,XXX) | 0.XXX |
| Emergency department | $X,XXX (X,XXX) | $X,XXX (X,XXX) | 0.XXX |
| Outpatient office | $X,XXX (X,XXX) | $X,XXX (X,XXX) | 0.XXX |
| Other outpatient | $X,XXX (X,XXX) | $X,XXX (X,XXX) | 0.XXX |
| Total pharmacy costs | |||
| Mean (SD) | $X,XXX (X,XXX) | $X,XXX (X,XXX) | 0.XXX |
| PPPM, mean (SD) | $XXX (XXX) | $XXX (XXX) | 0.XXX |
a Costs are health-plan-paid amounts (payer perspective), adjusted to 2025 USD using the medical care component of the Consumer Price Index (CPI).
b PPPM = per patient per month, normalized for variable follow-up time.
Abbreviations: IQR, interquartile range; PPPM, per patient per month; SD, standard deviation.
Shell 07 / 09
Table 7. Disease-specific healthcare costs during follow-up
Purpose: isolate cost burden directly attributable to the disease of interest, using the same category structure as the all-cause cost table for direct comparison.
| Disease-specific cost category, USD | Cohort A (N=X,XXX) | Cohort B (N=X,XXX) | P value |
|---|---|---|---|
| Disease-specific total costs | |||
| Mean (SD) | $X,XXX (X,XXX) | $X,XXX (X,XXX) | 0.XXX |
| Median (IQR) | $XXX (XXX–X,XXX) | $XXX (XXX–X,XXX) | 0.XXX |
| PPPM, mean (SD) | $XXX (XXX) | $XXX (XXX) | 0.XXX |
| Disease-specific medical costs | |||
| Mean (SD) | $X,XXX (X,XXX) | $X,XXX (X,XXX) | 0.XXX |
| PPPM, mean (SD) | $XXX (XXX) | $XXX (XXX) | 0.XXX |
| Inpatient | $X,XXX (X,XXX) | $X,XXX (X,XXX) | 0.XXX |
| Emergency department | $XXX (XXX) | $XXX (XXX) | 0.XXX |
| Outpatient office | $XXX (XXX) | $XXX (XXX) | 0.XXX |
| Other outpatient | $XXX (XXX) | $XXX (XXX) | 0.XXX |
| Disease-specific pharmacy costs | |||
| Mean (SD) | $XXX (XXX) | $XXX (XXX) | 0.XXX |
| PPPM, mean (SD) | $XX (XX) | $XX (XX) | 0.XXX |
a Costs are health-plan-paid amounts, adjusted to 2025 USD via the medical care CPI, reported from a payer perspective.
b Disease-specific is defined as in Table 5: a qualifying diagnosis in the primary position (sensitivity: any position); disease-specific pharmacy costs are limited to the defined drug list.
Abbreviations: IQR, interquartile range; PPPM, per patient per month; SD, standard deviation.
Shell 08 / 09
Table 8. Adjusted cost and HCRU analyses
Purpose: report multivariable-adjusted cost comparisons between cohorts, using a generalized linear model (GLM) for the full sample and a two-part model to separately handle the zero-cost mass.
Panel A · GLM (gamma distribution, log link)a
| Outcome | Cost ratio (exp β) | 95% CI | P value |
|---|---|---|---|
| Total costs | X.XX | X.XX–X.XX | 0.XXX |
| Medical costs | X.XX | X.XX–X.XX | 0.XXX |
| Pharmacy costs | X.XX | X.XX–X.XX | 0.XXX |
Panel B · Two-part model (≥X% of patients have zero cost)b
| Model part / outcome | Estimate | 95% CI | P value |
|---|---|---|---|
| Part 1 · Logistic (any cost vs. none) | OR | ||
| Cohort A vs. Cohort B (ref) | X.XX | X.XX–X.XX | 0.XXX |
| Part 2 · Gamma, log link (among patients with cost>0) | Cost ratio | ||
| Cohort A vs. Cohort B (ref) | X.XX | X.XX–X.XX | 0.XXX |
Combined predicted adjusted PPPMc
| Cohort | Predicted adjusted PPPM, USD | 95% CI |
|---|---|---|
| Cohort A | $X,XXX | $X,XXX–X,XXX |
| Cohort B | $X,XXX | $X,XXX–X,XXX |
| Difference (A–B) | $XXX | $XXX–XXX |
a Distribution family and link function selected via the modified Park test; the two-part model in Panel B is used when the zero-cost mass is too large for a single GLM to fit well.
b Covariates in both models: age, sex, US region, payer type, index year, baseline CCI, and baseline all-cause costs.
c Predicted adjusted PPPM combines Part 1 and Part 2 via recycled predictions (margins); 95% CIs computed via bootstrap or the delta method.
Abbreviations: GLM, generalized linear model; OR, odds ratio; CI, confidence interval; PPPM, per patient per month; CCI, Charlson Comorbidity Index.
Shell 09 / 09
Table 9 & Figure 1. Survival analysis shells
Purpose: report time-to-event results (Kaplan–Meier and Cox proportional hazards) for the pre-specified outcome, both as a summary table and as the accompanying survival curve figure.
Table 9 · Kaplan–Meier and Cox results
| Statistic | Cohort A (N=X,XXX) | Cohort B (N=X,XXX) |
|---|---|---|
| Events / censored, n (%) | XXX / XXX (XX.X) | XXX / XXX (XX.X) |
| Median time-to-event, months (95% CI) | XX.X (XX.X–XX.X) | XX.X (XX.X–XX.X) |
| Event-free probability, % (95% CI) | ||
| At 6 months | XX.X (XX.X–XX.X) | XX.X (XX.X–XX.X) |
| At 12 months | XX.X (XX.X–XX.X) | XX.X (XX.X–XX.X) |
| At 24 months | XX.X (XX.X–XX.X) | XX.X (XX.X–XX.X) |
| Unadjusted HR (95% CI)a | X.XX (X.XX–X.XX), P = 0.XXX | |
| Adjusted HR (95% CI)b | X.XX (X.XX–X.XX), P = 0.XXX | |
a Cohort B is the reference group; unadjusted comparison via the log-rank test.
b Adjusted for age, sex, US region, payer type, index year, baseline CCI, and baseline all-cause costs.
Abbreviations: HR, hazard ratio; CI, confidence interval; CCI, Charlson Comorbidity Index.
Figure 1 · Kaplan–Meier survival curves
| Numbers at risk | 0 | 5 | 10 | 15 | 20 | 25 | 30 |
|---|---|---|---|---|---|---|---|
| Cohort A | XXX | XXX | XXX | XXX | XXX | XXX | XXX |
| Cohort B | XXX | XXX | XXX | XXX | XXX | XXX | XXX |
a Censoring at disenrollment or end of data availability, whichever occurs first.
b Unadjusted between-cohort comparison via the log-rank test.
c The proportional hazards (PH) assumption is checked via Schoenfeld residuals; if PH does not hold, a landmark analysis or a time-varying-coefficient Cox model is substituted.
Abbreviations: KM, Kaplan–Meier; HR, hazard ratio; PH, proportional hazards; CI, confidence interval.
WORD TEMPLATE
Download the Word template
All nine shells — attrition, baseline demographics, baseline clinical characteristics, all-cause and disease-specific HCRU and cost, adjusted models, and the survival analysis shell — in one .docx file, formatted and ready to paste into a protocol or SAP.
Table shells only carry you as far as the SAP that specifies them. For the field-level definitions behind these
shells, see the repository; for claim-level source data,
see the annotated claim forms.