Certified Health Data Analyst (CHDA): Applied Preparation Guide Study Guide

Apply it

A clinic records 900 visits from 600 people during a month. Use 900 to describe visit workload and 600 to describe people reached. Document the counting unit before comparing the clinic with another site.

Apply it

A team wants to measure whether patients understood discharge instructions. Billing data can identify eligible encounters, but a suitably designed patient survey addresses understanding more directly. Combine sources only after defining eligibility and linkage rules.

Apply it

In a hypothetical inpatient surgical encounter, SNOMED CT can represent the documented clinical disorder and procedure. ICD-10-CM represents reportable diagnoses, ICD-10-PCS represents the hospital inpatient procedure, and CPT represents applicable professional services. A laboratory observation can use LOINC to identify the test, with its result and units recorded separately. A medication can use an RxNorm identifier. An analyst linking these records should preserve each representation's role rather than count every associated code as a separate clinical event.

Apply it

A manager compares October services with October cash receipts and concludes that collection performance fell. Reframe the analysis by following a defined service cohort through payment, or explicitly label the report as cash received during October.

Apply it

A hospital uses a CMS-approved hospital accreditation program to demonstrate compliance with applicable Medicare health and safety conditions. For a separate measure-submission obligation, the analyst still checks that program's population, exclusions, event definitions, deadlines, and validation requirements before reusing an internal dashboard.

Apply it

Among 400 initially unaffected people followed for one year, 20 develop the condition. Incidence proportion is 20/400, or 5%. A separate count of all existing cases on the year's final day answers a prevalence question.

Apply it

A hospital's documented screening rate rises after a new required EHR field appears. Examine the workflow change, field definitions, and prior documentation locations before concluding that screening itself increased by the same amount across the hospital.

Apply it

A clinic first counts missed appointments, then investigates scheduling patterns, then forecasts next week's absences. Selecting an outreach allocation under a fixed staffing budget adds a prescriptive decision to the earlier descriptive, diagnostic, and predictive work.

Apply it

A team proposes both extracting referral reasons from notes and predicting missed visits. Specify separate reference outcomes and evaluation plans. For the prediction task, exclude information recorded only after the appointment has already occurred.

Apply it

A director requests a spreadsheet of all appointments. Clarification shows the real decision is where to add reminder calls next month. Define the eligible appointment population, missed-visit measure, relevant groups, and resource constraints before designing the extract.

Apply it

For a referral-delay analysis, involve the service manager, scheduling staff, referral-system owner, and intended report users. Scheduling staff may reveal that a seemingly completed referral status actually means that an appointment was offered, not attended.

Apply it

A clinic defines its local missed-visit percentage as unattended appointments divided by eligible scheduled appointments, excluding documented cancellations before a stated cutoff. Record whether rescheduled appointments create new eligible records before comparing monthly performance.

Apply it

A clinic tests reminder calls to reduce missed appointments. Track call completion as a process measure, missed-visit percentage as the outcome, and staff time as a balancing measure. Increased calling alone does not establish improved attendance.

Apply it

Before comparing two clinics, document the eligible period, encounter types, measure rules, data cutoff, planned stratification, and validation totals. Obtain agreement on the reporting population before either clinic sees results that might influence its preferred definitions.

Apply it

Two days before delivery, a sponsor requests comparison with an external network. The analyst identifies new licensing, mapping, and comparability work, then proposes a revised deliverable or schedule instead of silently attaching unvalidated external figures.

Apply it

A dashboard deadline assumes next week's data access approval. While approval is pending, the team can finalize metric definitions and mock up displays, but it cannot honestly mark source validation complete or commit to final performance findings.

Apply it

A service dashboard is accepted only when its totals reconcile to the approved extract, its denominator is visible, and a manager can identify clinics exceeding the locally agreed review threshold without needing an analyst to reinterpret the display.

Apply it

Two dashboards report different discharge counts. Trace each value to its source table, status filters, cutoff time, and duplicate-handling rule. One may use finalized encounters while the other includes records still awaiting a status update.

Apply it

A hospital needs to know whether discharge instructions included a particular topic. If structured fields do not capture that detail, a defined chart-abstraction protocol may be more appropriate than inferring inclusion from a general discharge billing code.

Apply it

For a local April discharge analysis, select encounters using discharge time rather than admission time. Document the extraction cutoff so that later status corrections can be distinguished from errors when another analyst repeats the query.

Apply it

Every encounter has a discharge date, so the field is complete. A review finds some dates preceding admission. Add temporal validity checks and investigate source accuracy before using these records to calculate length of stay.

Apply it

One clinic lacks follow-up outcomes for 25% of eligible encounters while another lacks 2%. Compare recording workflows and characteristics of incomplete records before publishing a clinic ranking based only on observed follow-up outcomes.

Apply it

An extract contains two rows for one encounter because both diagnosis entries were joined. If the target dataset is encounter-level, correct the join or aggregate diagnoses appropriately. Do not remove legitimate encounters merely because the patient identifier repeats.

Apply it

Two sources encode appointment status as 'Completed' and 'COMP'. Map both to an approved shared category while preserving source labels. Route an unfamiliar code to review rather than automatically assigning it to the most common status.

Apply it

A medication map assigns most source codes to normalized concepts, but several combination products have multiple candidate matches. Hold those records in a review category and resolve the intended concept level before calculating ingredient utilization totals.

Apply it

Clinic A has 8 events among 100 encounters and Clinic B has 2 among 20. Combine the counts to obtain 10/120, or 8.33%, rather than averaging 8% and 10% without regard to volume.

Apply it

After mapping encounter statuses, total eligible encounters should remain unchanged under the approved rule. A 12% reduction prompts examination of unmatched codes and join behavior before the transformed file is released for analysis or reporting.

Apply it

To list every eligible encounter and any follow-up contact, place encounters on the left of a left join. Restrict qualifying contact types in the join condition when encounters without those contacts must remain in the output.

Apply it

One encounter has three diagnosis rows and two procedure rows. Joining both tables by encounter can produce six combinations. Aggregate each detail source to the required encounter-level features before calculating encounter-level totals across the cohort.

Apply it

A report needs clinics with at least 50 eligible May encounters. Filter May dates and eligible statuses with WHERE, group by clinic, and apply HAVING COUNT(*) >= 50 to retain groups meeting the volume requirement.

Apply it

A team measures the same employees' task completion times before and after training. Preserve the matched structure and analyze individual changes. Treating the two lists as unrelated groups discards information about how each employee changed.

Apply it

Five waiting times are 5, 6, 7, 8, and 74 minutes. The median is 7 minutes and the mean is 20 minutes. Report the median for the typical wait while investigating the long delay separately.

Apply it

Most visits last under two hours, but one appears to last 240 hours. Check whether the record represents a legitimate extended encounter, an incorrect end timestamp, or a unit conversion problem before changing or excluding it.

Apply it

For 25 matched records, the mean change is −4 minutes and the difference standard deviation is 10 minutes. The standard error is 10/√25 = 2 minutes, giving t = −2 against zero change and 24 degrees of freedom. Before using the small-sample t result, assess independence between pairs and examine the distribution of the 25 differences.

Apply it

A sample yields a mean wait of 30 minutes with a 95% confidence interval of 27–33 minutes. Describe uncertainty about the population mean; do not claim that 95% of individual patients waited between 27 and 33 minutes.

Apply it

Two independent samples have similar variability, but one contains 100 observations and the other 400. Expect the larger sample's standard error to be about half as large, without assuming that either sample represents every clinic equally well.

Apply it

A mean change has a 95% confidence interval of −6 to −1 minutes. Under the corresponding two-sided 5% test, zero change is rejected. The operational team must still decide whether the estimated reduction justifies implementation costs.

Apply it

A monthly event count doubles while eligible encounters also double and the event percentage stays constant. Report increased activity rather than a higher event risk, then check whether a newly connected clinic explains the larger reporting population.

Apply it

A clinic compares its all-visit measure with a network figure restricted to new patients. Recalculate a matched population or explain the mismatch before presenting the gap as evidence that one service performs better than the other.

Apply it

Consider hypothetical providers with identical follow-up and outcome definitions. Provider A has 18 outcomes among 900 patients younger than 65 and 10 among 100 patients aged 65 or older: age-specific rates are 2% and 10%, and the crude rate is 28/1,000 = 2.8%. Provider B has 2 outcomes among 100 younger patients and 90 among 900 older patients: the same age-specific rates, but a crude rate of 92/1,000 = 9.2%. Using a standard population weighted 50% to each age group, both adjusted rates equal (0.50 × 2%) + (0.50 × 10%) = 6%. Common weights remove the contribution of their different age distributions to this comparison.

Apply it

Before predicting missed appointments, remove fields created after the appointment outcome. Fit imputation and scaling using training records, then apply those fitted steps unchanged to held-out records that have not influenced model selection or preprocessing.

Apply it

An access project produces twenty charts, but the decision concerns staffing at two clinics. Lead with the clinics' comparable waiting-time results and demand volumes, then use other displays only where they explain the staffing choice or its uncertainty.

Apply it

A locally defined event percentage falls from 4% to 3%. Report a reduction of 1 percentage point and a relative reduction of 25%, together with eligible counts, so stakeholders can understand both the baseline and magnitude.

Apply it

Two clinics have estimated mean waits of 28 and 30 minutes, but the interval for their difference ranges from −5 to 9 minutes. Describe the comparison as uncertain rather than announcing a reliable two-minute advantage for either clinic.

Apply it

A report estimates average change using matched records with complete before-and-after measurements. Explain that the result describes those complete pairs and that excluded participants could affect interpretation if their missing measurements relate to their outcomes.

Apply it

A report includes outcomes for 700 of 1,000 eligible encounters. Show the observed outcome percentage, the 70% completeness, group-specific missingness, and the handling rule. Explain why the observed percentage may not represent every eligible encounter equally well.

Apply it

A patient-experience survey includes only respondents using an online portal. Describe the findings for that respondent group and examine coverage before claiming that the results represent all patients, including people without portal access or those who did not respond.

Apply it

Missed visits decline after reminder calls begin, but scheduling eligibility and available appointment times also change. Report the observed decline and concurrent changes; do not attribute the entire improvement to reminder calls without a design supporting that conclusion.

Apply it

A manager needs to review twelve monthly missed-visit percentages. Use a line chart with month on the horizontal axis, percentage on the vertical axis, and an annotation where the scheduling system changed the eligible appointment definition.

Apply it

Clinic event percentages are 4.0%, 4.2%, and 4.4%. A bar chart starting at 3.9% exaggerates the apparent differences. Use a zero-based bar scale or a clearly labeled dot display with suitable context and uncertainty.

Apply it

Replace 'FU compliance' with a locally accurate label such as 'Eligible discharges with documented follow-up within the specified window.' Add the period, denominator, source, and an explanation that the measure concerns documentation rather than confirmed patient understanding.

Apply it

An executive needs a service-level overview, while a clinic manager needs actionable local detail. Present the same approved measure definitions through appropriately aggregated views, keeping filter selections and reporting dates visible so users recognize which population they are seeing.

Apply it

A network's overall access measure improves, but subgroup reporting shows little change for one language group. Present both the total and stratified results, with group sizes and data limitations, so improvement planning can address the uneven pattern.

Apply it

One clinic has unusually long waits, but its timestamps may be recorded differently. Recommend validating the workflow and reviewing a targeted sample before reallocating staff, then specify which confirmed findings would justify testing a staffing change.

Apply it

An organization adopts a policy requiring approved access to analytics datasets. It assigns approvers, implements access requests, reviews permissions, and checks exceptions. Repeated inappropriate access indicates a need to improve implementation or controls rather than simply republish the policy.

Apply it

An encounter fact table has one row per encounter, clinic and date keys, and an event indicator. Clinic and date dimensions supply names and calendar groupings. Grouping by clinic and month supports encounter counts and event totals; calculate the event percentage from summed events divided by eligible encounters. Encounter counts do not automatically represent unique patients.

Apply it

A covered entity's analyst needs encounter dates and service categories for an internal operations report, with no applicable exception. Provide approved fields and identifiers only where necessary for that task instead of granting unrestricted access to complete clinical records.

Apply it

A covered entity creates a shared analytics workspace. It assigns individual accounts and role-appropriate permissions, records relevant activity, and defines procedures for access changes and incident handling rather than allowing analysts to share one unrestricted login.

Apply it

A dataset has names removed but retains full birth dates and detailed geographic information. Do not label it Safe Harbor de-identified on that basis. Evaluate the specified identifier requirements or obtain an appropriate Expert Determination for the proposed information.

Apply it

A submitted total is challenged. Review the activity record to identify the responsible run, then use the extract version, query version, reconciliation results, and approval record to determine how the figure was produced and whether correction is required.