Patient Appointments in Ambulatory Care

Predictive Optimization of Patient No-Show Management in Primary Healthcare Using Machine Learning

The "no-show" problem in healthcare refers to the prevalent phenomenon where patients schedule appointments with healthcare providers but fail to attend them without prior cancellation or rescheduling. In addressing this issue, our study delves into a multivariate analysis over a five-year period involving 21,969 patients. Our study introduces a predictive model framework that offers a holistic approach to managing the no-show problem in healthcare, incorporating elements into the objective function that address not only the accurate prediction of no-shows but also the management of service capacity, overbooking, and idle resource allocation resulting from mispredictions. Our approach simplifies preprocessing and eliminates the need for expert judgment in variable selection, thereby enhancing the model's usability in routine healthcare operations. Our research revealed that key predictors of no-shows are consistent across various studies. We employed semi-automatic feature selection techniques, achieving results comparable to state-of-the-art approaches but with significantly reduced complexity in their selection. This method not only streamlines the feature selection process but also enhances the overall efficiency and scalability of our predictive models, making them more adaptable to diverse healthcare settings. This comprehensive strategy enables healthcare providers to optimize resource allocation and improve service delivery, making our findings relevant for healthcare systems globally facing similar challenges. Future work aims to expand the analysis by incorporating additional third-party data sources, such as weather and commuting activities, to explore the broader impacts of external factors on patient no-show behavior. To the best of our knowledge, this innovative approach is expected to provide deeper insights and further enhance the predictability and effectiveness of no-show mitigation strategies in healthcare systems.

Predicting no-show appointments in a pediatric hospital in Chile using machine learning

The Chilean public health system serves 74% of the country's population, and 19% of medical appointments are missed on average because of no-shows. The national goal is 15%, which coincides with the average no-show rate reported in the private healthcare system. Our case study, Doctor Luis Calvo Mackenna Hospital, is a public high-complexity pediatric hospital and teaching center in Santiago, Chile. Historically, it has had high no-show rates, up to 29% in certain medical specialties. Using machine learning algorithms to predict no-shows of pediatric patients in terms of demographic, social, and historical variables. To propose and evaluate metrics to assess these models, accounting for the cost-effective impact of possible intervention strategies to reduce no-shows. We analyze the relationship between a no-show and demographic, social, and historical variables, between 2015 and 2018, through the following traditional machine learning algorithms: Random Forest, Logistic Regression, Support Vector Machines, AdaBoost and algorithms to alleviate the problem of class imbalance, such as RUS Boost, Balanced Random Forest, Balanced Bagging and Easy Ensemble. These class imbalances arise from the relatively low number of no-shows to the total number of appointments. Instead of the default thresholds used by each method, we computed alternative ones via the minimization of a weighted average of type I and II errors based on cost-effectiveness criteria. 20.4% of the 395,963 appointments considered presented no-shows, with ophthalmology showing the highest rate among specialties at 29.1%. Patients in the most deprived socioeconomic group according to their insurance type and commune of residence and those in their second infancy had the highest no-show rate. The history of non-attendance is strongly related to future no-shows. An 8-week experimental design measured a decrease in no-shows of 10.3 percentage points when using our reminder strategy compared to a control group. Among the variables analyzed, those related to patients' historical behavior, the reservation delay from the creation of the appointment, and variables that can be associated with the most disadvantaged socioeconomic group, are the most relevant to predict a no-show. Moreover, the introduction of new cost-effective metrics significantly impacts the validity of our prediction models. Using a prototype to call patients with the highest risk of no-shows resulted in a noticeable decrease in the overall no-show rate.

Using business intelligence and data visualization to understand the characteristics of failed appointments in dental school clinics

PURPOSE/OBJECTIVES

Broken appointments are an important cause of waste in health care. Patients who fail to attend incur costs to providers, deny trainees learning opportunities, and impact their own health as well as that of other patients who are waiting for care.

METHODS

A total of 410,000 appointment records over 3 years were extracted from our electronic health record. We conducted exploratory data analysis and assessed correlations between appointment no-shows and other attributes of the appointment and the patient. The University of Michigan Medical School's Committee on Human Research reviewed the study and deemed that no Institutional Review Board oversight was necessary for this quality improvement project that was, retrospectively, turned into a study with previously de-identified data.

RESULTS

The patient's previous attendance record is the single most significant correlation with attendance. We found that patients who said they are "scared" of dental visits were 62% as likely to attend as someone reporting "no problem." Patients over 65 years of age have better attendance rates. There was a positive association between receiving email/text confirmation and attendance. A total of 94.9% of those emailed a reminder and 92.2% of those who were texted attended their appointment.

CONCLUSION(S)

We were able to identify relationships of several variables to failed and attended appointments that we were previously unknown to us. This knowledge enabled us to implement interventions to support better attendance at Dental Clinics at the University of Michigan, improving patient health, student training, and efficient use of resources.

Fine for non-attendance in public hospitals in Denmark: A survey of non-attenders' reasons and attitudes

OBJECTIVE

To investigate non-attending patients' reasons for non-attendance and their general and specific attitudes towards a non-attendance fine.

DATA SOURCES

Non-attenders at two hospital departments participating in a trial of fine for non-attendance from May 2015 to January 2017.

DESIGN

A quantitative questionnaire study was conducted among non-attenders.

DATA COLLECTION

Non-attending patients in the intervention group were invited to complete the questionnaire. The response rate was 39% and the total number of respondents was 71 individuals.

PRINCIPAL FINDINGS

The main reason for non-attendance was technical challenges with the digital appointment and with cancelation. The main part of the respondents was generally positive towards a fine for non-attendance. However, approximately the half had a negative attitude towards the actual fine issued.

CONCLUSIONS

Technical challenges with appointments and cancelation should get special attention when addressing non-attendance. Danish non-attending patients are primarily positive towards the general principle of issuing a fine for non-attendance. However, a significant proportion of the generally positive, reported a negative specific attitude to the specific fine issued to them. This, however, did not affect their general attitude.

Predicting Patient No-show Behavior: a Study in a Bariatric Clinic

PURPOSE

No-shows of patients to their scheduled appointments have a significant impact on healthcare systems, including lower clinical efficiency and higher costs. The purpose of this study was to investigate the factors associated with patient no-shows in a bariatric surgery clinic.

MATERIALS AND METHODS

We performed a retrospective study of 13,230 records for 2660 patients in a clinic located in Rio de Janeiro, Brazil, over a 17-month period (January 2015-May 2016). Logistic regression analyses were conducted to explore and model the influence of certain variables on no-show rates. This work also developed a predictive model stratified for each medical specialty.

RESULTS

The overall proportion of no-shows was 21.9%. According to multiple logistic regression, there is a significant association between the patient no-shows and eight variables examined. This association revealed a pattern in the increase of patient no-shows: appointment in the later hours of the day, appointments not in the summer months, post-surgery appointment, high lead time, higher no-show history, fewer numbers of previous appointments, home address 20 to 50 km away from the clinic, or scheduled for another specialty other than a bariatric surgeon. Age group, forms of payment, gender, and weekday were not significant predictors. Predictive models were developed with an accuracy of 71%.

CONCLUSION

Understanding the characteristics of patient no-shows allows making improvements in management practice, and the predictive models can be incorporated into the clinic dynamic scheduling system, allowing the use of a new appointment policy that takes into account each patient's no-show probability.

A decision support system for real-time scheduling of multiple patient classes in outpatient services

We propose a methodology to provide real-time assistance for outpatient scheduling involving multiple patient types. Schedulers are shown how each prospective placement in the appointment book would impact a day's operational performance for patients and providers. Rooted in prior literature and analytical findings, the information provided to schedulers about vacant slots is based on the probabilities that the calling patient, the already-existing appointments, and the session-end time will be unduly delayed. The information is updated in real-time before and after every new booking; calculations are driven by each patient type's historical consultation times and no-show data, and implemented via a simulation tool based on the underlying analytical methodology. Our findings lead to practical guidelines for dynamically constructing templates that provide allowances for different consultation durations, service time variability, no-show rates, and provider-driven performance targets for patient waiting and provider overtime. Extensions to healthcare batch scheduling applications such as radiology, surgery, or chemotherapy-where patient mixes may be known in advance-are suggested as future research opportunities since avoiding session overtime and procedures' completion time delays involve similar considerations.