From Why Don't They Come to a Question With Variables: Two-Way Text Reminders and Missed First Appointments at a Behavioral Health Intake Clinic
Student Name
American College of Education
RES5303: Research Methods and Applied Statistics in Healthcare
Module 1 Assignment
Instructor Name
September 6, 2027
The Broad Interest
The intake clinic of a community behavioral health organization, a composite invented for this assignment, schedules about 150 first appointments a month for adults referred for depression, anxiety and substance use concerns. Over the past year, 31 percent of those first appointments were missed without cancellation. The clinic manager's question is a natural one: why don't they come? It is a question worth asking and impossible to answer as asked. It has no defined population, no measurable outcome and no single explanation that could be tested, and the possible answers range from transportation and work schedules to the stigma of seeking mental health care.
Missed appointments carry real costs. A study of a large health care system found a mean no-show rate of 18.8 percent across its main clinics, estimated an average cost of $196 per missed appointment and found that a centralized telephone reminder system reduced the rate only modestly (Kheirkhah et al., 2016). In behavioral health, the stakes are higher still, because a missed first appointment often means a person who reached out for help at a difficult moment is lost to care entirely. Mitchell and Selmes (2007) describe missed appointments in psychiatric services as a signal of disengagement that is associated with worse outcomes, and they note that reasons range from forgetting to ambivalence about treatment. A research question has to choose one of those reasons and make it measurable, or it will never produce an answer the clinic can use.
What Is Already Known
Reminders are the most studied response to missed appointments. Guy et al. (2012) conducted a meta-analysis of text message reminders and clinic attendance. Across eight randomized trials, text reminders increased the odds of attendance, with a summary odds ratio of 1.48, and the effect did not differ significantly by clinic type, by how far in advance the message was sent or by patient age. The observational studies in the review were too inconsistent to combine.
Two gaps in that evidence matter for this clinic. The trials mostly tested one-way reminders, messages that inform but do not invite a reply. A two-way reminder, which asks the patient to confirm, cancel or reschedule by replying, might work differently, both by prompting a decision and by freeing slots that can be offered to others. And few of the trials were conducted in behavioral health intake, where the first appointment is often scheduled days or weeks after a referral made at a moment of distress. A question focused on two-way reminders for first behavioral health appointments would add to what is known rather than repeat it.
Naming the Variables
The independent variable is the type of reminder the patient receives before the first appointment. It has three categories: no text reminder, the clinic's current practice for patients without a mobile number on file; a one-way text reminder sent two days before the appointment; and a two-way text reminder sent at the same time that asks the patient to reply C to confirm, R to reschedule or X to cancel. It is a nominal variable, drawn from the reminder system's message log.
The dependent variable is attendance at the first scheduled appointment, defined as the patient arriving and being seen on the scheduled date, recorded in the scheduling system as completed. Appointments canceled more than 24 hours in advance are excluded from the denominator, because the clinic can reuse those slots; late cancellations and no-shows are counted as not attended. It is a dichotomous nominal variable. A secondary dependent variable, the proportion of freed slots filled by another patient, is included because a two-way reminder could improve the clinic's use of time even if attendance did not change.
Several variables could confound the relationship, because they affect both whether a patient receives a text reminder and whether the patient attends. They include lead time, the number of days between referral and the scheduled appointment, a ratio variable; age in years; insurance type, a nominal variable with four categories; referral source, a nominal variable separating self-referrals, primary care referrals and emergency department referrals; and the presenting concern recorded at referral. Each is available in the scheduling or referral record. Naming the confounders in advance is what separates a question about an association from a question that can be answered fairly.
The Question and Hypothesis
The research question is: among adults referred for a first appointment at a community behavioral health intake clinic, is the type of text reminder received, none, one-way or two-way, associated with attendance at the first scheduled appointment, after accounting for lead time, age, insurance type, referral source and presenting concern? The directional hypothesis is that patients who receive a two-way reminder will have higher attendance than those who receive a one-way reminder, who in turn will have higher attendance than those who receive none.
The question as written calls for an analytic design. If the clinic can assign reminder type at random, a randomized comparison would answer it most directly and would remove confounding by design. If it cannot, a retrospective cohort comparison using existing records would be possible, with the named confounders controlled statistically. The choice between those designs, and what the resulting sample can honestly represent, are the subjects of the next two modules.
What the Question Leaves Out
A well-built question is narrow on purpose, and this one excludes several things the manager cares about. It does not ask why patients miss appointments, which would require interviews or surveys and a qualitative or mixed design; it asks whether one changeable factor, the reminder, is associated with attendance. It does not address attendance at follow-up visits after the first appointment, where the reasons for disengagement are likely to differ. And it does not measure clinical outcomes such as symptom change, which would require following patients over months and linking to data the clinic does not routinely collect.
The question also carries assumptions that later modules must test. It assumes the reminder system logs every message reliably, including failed deliveries, which the informatics team will need to confirm. It assumes the scheduling system records late cancellations consistently rather than as no-shows, which a sample of records can check. And it assumes that patients without a mobile number, who form the no-reminder group, are not so different from others that no amount of statistical adjustment could make the comparison fair. That last assumption is the strongest reason to prefer random assignment of reminder type among patients who do have mobile numbers, a design the next modules will consider. A question is only as sound as the data that will answer it, so its assumptions belong on the page alongside its variables.
Conclusion
The manager's broad question, why don't they come, has become a question with named variables: reminder type as the independent variable in three categories, attendance at the first appointment as the dependent variable with a clear rule for cancellations, and five potential confounders, each defined and tied to a data source. The question builds on meta-analytic evidence that text reminders improve attendance while targeting two gaps in that evidence, two-way messaging and behavioral health intake. It is now specific enough for a design to be chosen, a sample to be drawn and an answer to be found.
References
Guy, R., Hocking, J., Wand, H., Stott, S., Ali, H., & Kaldor, J. (2012). How effective are short message service reminders at increasing clinic attendance? A meta-analysis and systematic review. Health Services Research, 47(2), 614-632. https://doi.org/10.1111/j.1475-6773.2011.01342.x
Kheirkhah, P., Feng, Q., Travis, L. M., Tavakoli-Tabasi, S., & Sharafkhaneh, A. (2016). Prevalence, predictors and economic consequences of no-shows. BMC Health Services Research, 16, Article 13. https://doi.org/10.1186/s12913-015-1243-z
Mitchell, A. J., & Selmes, T. (2007). Why don't patients attend their appointments? Maintaining engagement with psychiatric services. Advances in Psychiatric Treatment, 13(6), 423-434. https://doi.org/10.1192/apt.bp.106.003202
How this RES 5303 Module 1 example is structured
RES 5303 Module 1 often turns a broad healthcare interest into a question with named variables; your classroom's instructions decide whether the deliverable is a short paper, a worksheet or both. This example describes the interest and the local figure behind it, reviews what is already known so the question adds something, and then builds the question one variable at a time: independent, dependent and the variables that could confound the relationship. Each variable gets an operational definition, a level of measurement and a data source. The paper ends with the question, a hypothesis and what kind of design the question calls for, which sets up the modules on sampling and design.
RES5303 Module 1 questions, answered
What does RES5303 Module 1 usually ask for?
RES5303 Module 1 often asks students to take a broad healthcare interest and turn it into a researchable question with named variables. Many sections expect independent and dependent variables to be identified, defined and classified by level of measurement. Your classroom's instructions decide the format, which may be a paper, a worksheet or a discussion.
What is an operational definition?
An operational definition states exactly how a variable will be measured in your study, including the data source and any rules for edge cases. Attendance, for example, might be defined as being seen on the scheduled date, with cancellations more than 24 hours ahead excluded. A clear operational definition is what makes a variable measurable.
Why identify confounding variables in a research question?
A confounder affects both the independent and the dependent variable, so it can create or hide an association. Naming confounders in advance lets you choose a design or analysis that accounts for them, such as randomization or statistical adjustment. Leaving them out makes any result hard to interpret.
Write yours, or have the desk draft it
This paper is an original model document written by our desk, not a submitted student paper and not an official American College of Education document. Read it for the moves, then write your own to the instructions in your classroom. If you want one built to your exact prompt and rubric, the first custom sample is free and arrives in 24 to 48 hours.