DDHA 8303 Week 7 Sampling Plan Example

Every Coder on the Roster, Sixteen in the Interview Room and the Ones Who Already Left: A Sampling Plan for a Study of Remote Inpatient Coders

Student Name

Walden University

DDHA 8303: Research Methods for Healthcare Administrators

Instructor Name

Month Day, Year

Every Coder on the Roster, Sixteen in the Interview Room and the Ones Who Already Left: A Sampling Plan for a Study of Remote Inpatient Coders

Target Population

The study's target population is every person employed by Halvorsen Health as an inpatient coder at any time from January 2023 through December 2025. For this plan, an inpatient coder is anyone who held the human resources job code for Inpatient Coder I or Inpatient Coder II and who assigned DRGs to inpatient accounts during that period. The definition is written so that an administrator could apply it to a roster without judgment: role by job code, unit by the inpatient coding department, tenure by dates in the job code and exposure by the work arrangement recorded in each month. Coding supervisors, outpatient and professional fee coders and clinical documentation specialists fall outside it, even if they occasionally coded inpatient accounts, because their work is audited differently.

Sampling Frame

The frame is built from three sources: the human resources roster of job codes, the coding workflow system's list of users who finalized inpatient accounts and the internal audit database. Matching the three produces 78 people who meet the definition, of whom 64 were employed at the end of 2025 and 14 had left.

The frame is not identical to the population, and the differences matter. First, agency coders hired through a staffing firm to clear backlogs appear in the workflow system but not in the roster, so their work arrangements are not recorded; there were nine during the period. Second, two coders changed job codes mid-period and appear twice under different identifiers, which the matching resolved by employee number. Third, the audit database begins recording coder identifiers consistently only from January 2023, which is why the study period starts there. Everyone who qualifies under the definition and was employed by Halvorsen can be listed. Agency coders qualify in every respect except employment and cannot be listed with the information the study needs.

Selection Technique

The two branches of the study select differently, consistent with the design argued in Week 5.

The quantitative branch uses a census. Every person in the frame with at least six consecutive months of audit records is included, because the difference-in-differences analysis depends on comparing each coder with themselves over time and on comparing groups that moved at different times. Sampling from 78 people would add sampling error without saving meaningful effort, since the records already exist.

The qualitative branch uses purposeful sampling for maximum variation. Palinkas et al. (2015) describe purposeful sampling as selecting information-rich cases, and maximum variation as choosing participants who differ on dimensions expected to shape the phenomenon. Here, those dimensions are work arrangement, move wave, tenure and hospital of origin. The target is sixteen coders: six from the January 2024 wave, five from the September wave and five who stayed onsite or hybrid, with each group including coders above and below five years of tenure and from at least three of the five hospitals.

Recruitment

Recruitment for the interviews proceeds in five steps, each dependent on a specific kind of access.

In the first step, the system's research office, not the director of health information management, sends an invitation to every current coder in the frame. This depends on the research office's agreement and on the coders' work email list, which human resources has agreed to provide. Second, coders who are interested reply directly to the research office, which keeps their response private from coding management. Third, a member of the research team screens respondents against the variation targets, using arrangement, wave, tenure and hospital from the frame rather than asking respondents to report them. Fourth, selected coders are scheduled for a 45-minute interview by video or in person during paid time that is excluded from productivity tracking, which depends on the coding manager's written agreement to suspend the productivity count for that period. Fifth, the consent process described in the ethics protocol takes place before the interview begins.

The quantitative branch requires no recruitment, because it uses existing audit records under an approved data use agreement.

Inclusion and Exclusion Rules

Inclusion for the quantitative branch: listed in the frame; at least six consecutive months with ten or more audited accounts; work arrangement recorded for every included month.

Inclusion for the qualitative branch: all of the above, plus currently employed at Halvorsen and holding the inpatient coder job code on the date of invitation.

Exclusion for both branches: any month in which the coder was in orientation, on leave for more than half the month or assigned to a special project that removed them from routine coding, as recorded in human resources and workflow data.

Exclusion for the qualitative branch only: members of the research team's own reporting line. One coder who serves on the audit team that produces the outcome data is excluded from interviews to avoid confusing her two roles.

Each rule can be checked against a record. None relies on a judgment about the coder's suitability.

Coverage: Who the Plan Misses

The most important group the plan cannot fully reach is the fourteen coders who left. Their audit records remain, so the quantitative branch includes them up to their departure. The interviews cannot include them. If coders who struggled with remote work were more likely to leave, the interviews will hear mainly from those for whom it worked, and the qualitative account of how remote coders resolve uncertain decisions will be biased toward successful strategies. The plan addresses this by asking human resources whether exit interview summaries for those fourteen can be shared in de-identified form; if they can, they will be analyzed as a separate source and labeled as such.

The second group is the nine agency coders. They did coding work during the period but cannot be placed in a work arrangement group. Excluding them is necessary, but readers should know that some accounts coded during busy months were coded by people outside the study.

The third group is current coders who decline interviews. Groves and Peytcheva (2008), in a meta-analysis of survey nonresponse, showed that a low response rate by itself says little about how biased results will be; what matters is whether the reasons people decline are connected to what is being studied. For this study, a coder anxious about their accuracy might be less willing to talk, which would be such a relationship. The plan will compare interviewed and non-interviewed coders on tenure, arrangement and audit agreement, using frame data, and report any differences.

What This Means for the Claim

The quantitative findings will apply to Halvorsen's employed inpatient coders over the study period, including those who left. The qualitative findings will describe how current coders, and possibly some former ones through exit data, explain their practices, and they should not be read as representing everyone who tried remote coding. Polit and Beck (2010) suggest that qualitative findings support transferability when their conditions are described clearly enough for readers to judge similarity. This plan's coverage section is part of that description.

References

Groves, R. M., & Peytcheva, E. (2008). The impact of nonresponse rates on nonresponse bias: A meta-analysis. Public Opinion Quarterly, 72(2), 167-189. https://doi.org/10.1093/poq/nfn011

Palinkas, L. A., Horwitz, S. M., Green, C. A., Wisdom, J. P., Duan, N., & Hoagwood, K. (2015). Purposeful sampling for qualitative data collection and analysis in mixed method implementation research. Administration and Policy in Mental Health and Mental Health Services Research, 42(5), 533-544. https://doi.org/10.1007/s10488-013-0528-y

Polit, D. F., & Beck, C. T. (2010). Generalization in quantitative and qualitative research: Myths and strategies. International Journal of Nursing Studies, 47(11), 1451-1458. https://doi.org/10.1016/j.ijnurstu.2010.06.004