Parente et al.

BMC Health Services Research (2018) 18:185

https://doi.org/10.1186/s12913-018-2979-z

RESEARCH ARTICLE Open Access

Using overbooking to manage no-shows in

an Italian healthcare center
Chiara Anna Parente1*, Domenico Salvatore2, Giampiero Maria Gallo3 and Fabrizio Cipollini4

Abstract
Background: In almost all healthcare systems, no-shows (scheduled appointments missed without any notice from
patients) have a negative impact on waiting lists, costs and resource utilization, impairing the quality and quantity
of cares that could be provided, as well as the revenues from the corresponding activity. Overbooking is a tool
healthcare providers can resort to reduce the impact of no-shows.
Methods: We develop an overbooking algorithm, and we assess its effectiveness using two methods: an analysis of
the data coming from a practical implementation in an healthcare center; a simulation experiment to check the
robustness and the potential of the strategy under different conditions. The data of the study, which includes personal
and administrative information of patients, together with their scheduled and attended examinations, was taken from
the electronic database of a big outpatient center. The attention was focused on the Magnetic Resonance (MR) ward
because it uses expensive equipment, its services need long execution times, and the center has actually used it to
implement an overbooking strategy aimed at reducing the impact of no-shows. We propose a statistical model for the
patient’s show/no-show behavior and we evaluate the ensuing overbooking procedure implemented in the MR ward.
Finally, a simulation study investigates the effects of the overbooking strategy under different scenarios.
Results: The first contribution is a list of variables to identify the factors performing the best to predict no-shows. We
classified the variables in three groups: “Patient’s intrinsic factors”, “Exogenous factors” and “Factors associated with the
examination”. The second contribution is a predictive model of no-shows, which is estimated on context-specific data
using the variables just discussed. Such a model represents a fundamental ingredient of the overbooking strategy we
propose to reduce the negative effects of no-shows. The third contribution is the assessment of that strategy by means
of a simulation study under different scenarios in terms of number of resources and no-show rates. The same
overbooking strategy was also implemented in practice (giving the opportunity to consider it as a quasi-experiment) to
reduce the negative impact caused by non attendance in the MR ward. Both the quasi-experiment and the simulation
study demonstrated that the strategy improved the center’s productivity and reduced idle time of resources, although
it increased slightly the patient’s waiting time and the staff’s overtime. This represents an evidence that overbooking
can be suitable to improve the management of healthcare centers without adversely affecting their costs and the
quality of cares offered.
Conclusions: We shown that a well designed overbooking procedure can improve the management of medical centers,
in terms of a significant increase of revenue, while keeping patient’s waiting time and overtime under control. This was
demonstrated by the results of a quasi-experiment (practical implementation of the strategy in the MR ward) and a
simulation study (under different scenarios). Such positive results took advantage from a predictive model of no-show
carefully designed around the medical center data.
Keywords: No-show, Overbooking, Healthcare, Logistic regression, Simulation, Scheduling

* Correspondence: [email protected]
1
IRCCS SDN, Napoli, Via E. Gianturco 113, 80143 Naples, Italy
Full list of author information is available at the end of the article

© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Parente et al. BMC Health Services Research (2018) 18:185 Page 2 of 12

Background patients and the costs of patients waiting time and staff
Patient’s non–attendance, defined as “missing a scheduled overtime [4, 14–16].
appointment without canceling it”, is an important issue This aim of this paper is to develop a context-specific
for the management of healthcare centers. Although re- overbooking algorithm and to assess its effectiveness
ported no-show rates vary widely, from a minimum of 3% using two methods: 1) the analysis of real data coming
to a maximum of 80% [1], depending on the type of facility from the implementation of the algorithm in an health-
and practice [2, 3], a patient’s non–attendance to scheduled care center and 2) a simulation experiment to examine
appointments may affect productivity, consume resources, the robustness and potential of the strategy under differ-
prolong the waiting time for an examination and reduce ent conditions.
customer satisfaction. These economic effects have direct At the best of our knowledge, our study is one of the
clinical counterparts, because poorly managed no-shows first proposing an algorithm for overbooking in health-
interrupt continuity and quality of cares [4], delay diagno- care centers, also implemented in practice. The only
ses and treatments of other patients and compromise the overbooking algorithm we found was proposed by [4].
intention of health companies to invest in new techno- Our algorithm is novel because:
logical and human resources.
A stream of research investigated the factors explain- 1. It considers different levels of no-show and a differ-
ing the propensity of patients to no-show. The list of the ent number of available MR scanners;
possible explanatory variables include: 2. It simulates the variables show/no-show, actual
duration and patient being early/late;
1 factors related to individuals, such as demographic 3. It is actually implemented in the MR ward of the
characteristics, ethnicity, type of medical insurance, healthcare center.
socioeconomic conditions, disease-related factors (i.e.
acute vs. chronic disease), the past history in terms of The healthcare center, located in Southern Italy is a
previous reservations, no-shows and cancellations [2]; leading provider of medical imaging and laboratory diag-
2 environmental conditions, such as office accessibility, nostic services. It operates within the Italian National
difficulties in reaching the healthcare center, lack of Health Service (NHS) and with health insurance com-
transportation, lack of child care, lead time of panies, offering many diagnostic procedures structured
examination, examination type and service quality in different wards.
[2, 5–7]; Italy’s healthcare system provides universal coverage
3 practices designed to increase patient’s attendance largely free of charge at the point of delivery. Public
rates, as education, penalties and reminders by funding is collected at the national level through general
phone or email [5, 8–11]. taxation and then distributed to 21 regions, which are in
charge of making most administrative decisions on the
Early works on patient’s non–attendance assumed organization of healthcare. In turn, regional administra-
homogeneous patients, i.e. sharing the same no-show tions deliver services within their area through health
probability [12]; more recent contributions allowed such a districts, which may be in charge of some public health
probability to be different across patients, so as to estimate decisions and collaborate with local municipalities on
it based on their specific characteristics and historical data health and public assistance matters.
[13–15]. Some appointment scheduling techniques Care is provided by public, not for profit and private
discussed in existing literature assumed a deterministic providers (such as the healthcare center in this study)
and fixed service time [4], while others considered it as accredited by the public health service which funded
random, for example exponentially [12, 14, 15] or gamma 78.2% of overall health-care spending in 2012 [17].
[13] distributed. These approaches aimed at identifying Public providers may be independent public hospitals or
which time of the day an examination should be booked, organizational units of health districts.
restricting the freedom of choice of the patients [14, 16]. The catchment area identifies the healthcare center as
Such scheduling procedures were frequently associated important in the region, as it includes patients from
with overbooking (defined as multiple-booking of patients mostly the province of location, but also a few from the
in a common time slot) aimed at increasing the number neighboring provinces and even from the neighboring
of patients the healthcare center could receive. As regions. The healthcare center only provides ambulatory
reported in [4], overbooking practices were more con- medical care diagnostic procedures and it does not pro-
venient when the no-show rate was high, the health- vide hospital (inpatient) care nor primary care services.
care center served a larger number of patients and the Patients have access to the procedures in the healthcare
service time variability was low, with the aim of finding center only if referred by a primary care doctor or a by a
an optimal compromise between expected additional specialist doctor; each referral should be accompanied
Parente et al. BMC Health Services Research (2018) 18:185 Page 3 of 12

by the medical indication for the diagnostic procedure what follows, we focused on the latter, because it has long
ordered. Some of the diagnostic procedures available for execution time and needs expensive equipment; moreover,
outpatients in this specific healthcare center may be com- since the healthcare center has later implemented an over-
plex and not available even in big secondary care hospitals. booking strategy, this gave us the opportunity to evaluate
the effectiveness of such a strategy in areal setting. Models
Methods estimated by ward differ just by the explanatory variables,
Data with MR and CT sharing the same list, while in the other
All the data in this study were taken from the elec- wards the Contrast Agent is excluded, as not needed.
tronic database of this big outpatient healthcare Moreover, Mammography did not include Gender and
center specialized in diagnostic procedures. The data- excluded the youngest 0–18 age group (one patient,
base includes personal and administrative information excluded in agreement with the recommendations that
of the patients together with all their scheduled and the execution of that examination “not before age 25”
attended examinations. [18]). Orthopantomography did not include the Price
The healthcare center accepts reservations through its of the Examination because there was a fixed price
call center or its website and charges the costs of the for all patients.
procedures to the patient only after the execution of the For the MR ward, descriptive statistics of the variables
examination and only in case there is any amount not considered are reported in Table 2 (outliers for variables
covered by the NHS or by patient’s insurance. Therefore, age and waiting time were removed).
there is no direct financial loss for a non-attending
patient. As noted, because of the regional regulation, the
A statistical model for show/no-show
healthcare center cannot perform the examinations
A logistic regression analysis [19] was used to build a pre-
considered in this study unless an external specialist
dictive model for a single examination’s no-show yi(i = 1,
medical doctor has prescribed it, in order to guarantee
…, N) conditional on a vector of explanatory variables xi,
the appropriateness of the procedure.
with clustered robust standard errors computed to correct
The reference population of this study included all
for repeated examinations on the same patient:
patients booking an examination from January 1, 2012
to December 31, 2014 for a total of 104,188 patients and
152,547 examinations. During this interval, the average yi j xi  Beðπ i Þ
rate of no-show was 14.6%; 73.8% of patients attended e ηi
πi ¼
their booked examinations, 16.2% were no-show only 1 þ eηi
once and the remaining 10% had more than one ηi ¼ β0 þ β1 x1;i þ ⋯ þ βk xk;i
no-show. The study included 45.8% males and 54.2%
females with an average age, respectively, of 52 and 53. where:

Analysis  Be(πi) is a Bernoulli distribution with no-show

The main purpose of this work is to develop and probability πi;
estimate a model to predict whether a booked examin-  β0, β1, …, βk are the regression coefficients;
ation may result in a no-show. This tool helps the center  x1, i, x2, i, …, xk, i are the explanatory variables (cf. 1).
in its overbooking strategy to reduce the fall in revenue
caused by non–attending patients, while not increasing
the patient’s waiting time and the overtime of medical An overbooking quasi experiment
staff. The potential explanatory variables (defined in To reduce the negative effects caused by non–attend-
Table 1 and homogeneously classified in Table 2) repre- ance, starting from January 1, 2015 the healthcare center
sent characteristics of the individuals (for example, their implemented an overbooking strategy in its MR ward.
age or a previous no-show), factors associated with the The reference population included all patients booking
examination (for example if it involved the use of a an examination from January 1, 2015 to June 31, 2015
contrast agent or its price) and exogenous factors (for for a total of 25,395 patients and 29,320 examinations.
example date and time of the appointment or the During this interval, the average rate of no-show was
weather forecast at that day). The base levels for the cat- 14.8%; 68.8% of patients attended all their booked examina-
egorical variables are reported in Table 3. We performed tions, 12.1% were no-show only once and the remaining
the analysis separately for each department of the health- 19.1% had more than one no-show. The study included
care center (Echography, Echo Doppler, Mammography, 47.1%, males and 52.9% females with an average age,
Orthopantomography, Radiography, Computer-Assisted respectively of 53 and 52.
Tomography – CT, and Magnetic Resonance - MR); in The correspondent algorithm worked as follows:
Parente et al. BMC Health Services Research (2018) 18:185 Page 4 of 12

Table 1 Definition of the variables included in the analyses

Variable Description
No-show Missing a scheduled appointment without canceling it. When a patient cancelled his reservation, in fact, this was
removed just from the dependent variable because a cancelled appointment could not be classified as a no-show
because the slot was made available for another patient.
Insurance In Italy, the majority of the examinations is covered by the NHS, but there is a portion that is privately paid by
status patients
• NHS: Examination covered by the NHS
• Private: Examination paid directly by the patient or by a private insurance
Rate of previous cancellations Number of cancellations divided by number of reservations in previous examinations of the booked patient since
2012 in any ward
Rate of previous no-shows Number of no-show divided by number of reservations in previous examinations of the booked patient since 2012
in any ward
Booking An automated phone service calls patients aged over 65 2 days before the appointment to get a confirmation
confirmation • Not Confirmed: if the patient did not confirm the appointment
• Confirmed: if the patient confirmed the appointment
Type of booking Method of reservation used by patients
• Phone: reservation by phone with the assistance of an operator
• Web: reservation through the institutional website (active since April 2014)
Time of the day MR and CT wards had an opening time ranging from 6 AM to 2 AM (+1d); all remaining wards had opening time
8 AM - 8 PM. Accordingly, we considered the following time of the day:
• 6 AM–8 AM: for MR and CT wards only
• 8 AM-1 PM: for all wards
• 1 PM–8 PM: for all wards
• 8 PM-2 AM (+1d): for MR and CT wards only
Long weekend It indicated whether the day of the appointment fell in a week with a public holiday
• Yes: week with a public holiday
• No: week without a public holiday
Weather forecast An automated system downloads every day from the web (http://api.wunderground.com) the weather forecasts
of the same and of the following 3 days
• Clear: day without rain or storm
• Rain: day with rain
• Storm: day with a storm
Text message reminder Two days before the examination, an automated service sends a text message to all patients aged between 18
service and 65 (active since April 2014)
• Not yet activated: examinations booked before April 2014
• Activated but not sent: examinations booked since April 2014 without sending a text message reminder
because the system did not know the patient mobile phone number or because he/she was over 65
• Sent: examination booked after April 2014 with text message reminder sent
No NHS coverage period In Italy, the NHS covers most of the examinations until a cap on the NHS budget was reached, causing a stop
in the coverage. This happened, usually, in the last weeks of the year
• No: examination booked in the coverage period
• Yes: examination booked when there was no NHS insurance because the budget was spent in full
Price of the examination Price paid by the patient. It was set to zero for the examinations covered by NHS and to the whole price
otherwise
Waiting list Number of days between the booking and the examination
Time allowed Expected duration of the examination
Hourly revenue The revenues of all examinations divided by the product between the number of working hours and the
number of active MR scanners
Waiting time Difference between the time when the patient starts his/her examination and the latest between the scheduled
time of the appointment and the arrival time at the healthcare center
Idle time Idle time of the scanners across the appointments
Overtime Staff overtime at the end of the day

1. The lower bound of the 95% Confidence Interval theoretical time duration of the booked examination.
(CI) of the no-show probability was computed (based The results of this product represented the number
on the coefficients of the model in section A statistical of expected minutes eligible for overbooking
model for show/no-show) and then multiplied by the appointments.
Parente et al. BMC Health Services Research (2018) 18:185 Page 5 of 12

Table 2 Descriptive statistics

Variable Total No-Show Show Test statistic
Patient’s intrinsic factors
Gender 20.56 ***
Male 45.78%(35,152) 45.78 (35152) 46.11%(30,374)
Female 54.22%(41,629) 56.23%(6137) 53.89%(35,492)
Age group 224.26 ***
0–18 4.26%(3268) 4.56%(498) 4.21%(2770)
19–45 35.2%(27,030) 40.35%(4404) 34.35%(22,626)
46–64 37.64%(28,901) 36.78%(4014) 37.78%(24,887)
65–79 20.31%(15,597) 15.9%(1736) 21.04%(13,861)
80+ 2.59%(1985) 2.41%(263) 2.61%(1722)
Insurance status 10.68 **
NHS 86.31%(66,270) 87.31%(9530) 86.14%(56,740)
Private 13.69%(10,511) 12.69%(1385) 13.86%(9126)
Rate of previous cancellations 0.21 ± 0.26 0.19 ± 0.25 0.21 ± 0.26 6.22 ***
Rate of previous no-shows 0.04 ± 0.11 0.05 ± 0.13 0.03 ± 0.1 −14.11 ***
Type of patient 443.76 ***
New patient 37.25%(28,602) 46.28%(5052) 35.75%(23,550)
Returned patient 62.75%(48,179) 53.72%(5863) 64.25%(42,316)
Booking confirmation 1294.47 ***
Not confirmed 58.61%(45,005) 74.33%(8113) 56.01%(36,892)
Confirmed 41.39%(31,776) 25.67%(2802) 43.99%(28,974)
Type of booking 42.05 ***
Contact center 99.8%(76,629) 99.54%(10,865) 99.85%(65,764)
Web 0.2%(152) 0.46%(50) 0.15%(102)
Exogenous factors
Day of the week 44.49 ***
Monday 15.61%(11,985) 16.76%(1829) 15.42%(10,156)
Tuesday 16.82%(12,918) 16.56%(1807) 16.87%(11,111)
Wednesday 16.19%(12,432) 15.72%(1716) 16.27%(10,716)
Thursday 16.61%(12,751) 16.68%(1821) 16.59%(10,930)
Friday 16.53%(12,694) 15.62%(1705) 16.68%(10,989)
Saturday 12.22%(9380) 11.64%(1270) 12.31%(8110)
Sunday 6.02%(4621) 7.03%(767) 5.85%(3854)
Month of the year 46.1 ***
January 9.72%(7466) 9.73%(1062) 9.72%(6404)
February 10.34%(7940) 10.93%(1193) 10.24%(6747)
March 11.34%(8709) 11.74%(1281) 11.28%(7428)
April 9.69%(7437) 9.69%(1058) 9.68%(6379)
May 11.23%(8619) 10.43%(1138) 11.36%(7481)
June 10.15%(7797) 10.46%(1142) 10.1%(6655)
July 9.74%(7478) 9.69%(1058) 9.75%(6420)
August 5.58%(4285) 5.34%(583) 5.62%(3702)
September 9.47%(7270) 8.84%(965) 9.57%(6305)
October 7.12%(5464) 7.95%(868) 6.98%(4596)
Parente et al. BMC Health Services Research (2018) 18:185 Page 6 of 12

Table 2 Descriptive statistics (Continued)

Variable Total No-Show Show Test statistic
November 3.55%(2726) 2.99%(326) 3.64%(2400)
December 2.07%(1590) 2.21%(241) 2.05%(1349)
Year 83.09 ***
2012 28.81%(22,122) 32.24%(3519) 28.24%(18,603)
2013 34.55%(26,530) 34.08%(3720) 34.63%(22,810)
2014 36.64%(28,129) 33.68%(3676) 37.13%(24,453)
Time of the day 186.99 ***
6 AM–8 AM 12.02%(9229) 11.66%(1273) 12.08%(7956)
8 AM-1 PM 35.92%(27,582) 31.54%(3443) 36.65%(24,139)
1 PM–8 PM 39.46%(30,299) 40.88%(4462) 39.23%(25,837)
8 PM-2 AM (+1d) 12.6%(9671) 15.91%(1737) 12.05%(7934)
Long weekend 0.69
No 85.64%(65,753) 85.9%(9376) 85.59%(56,377)
Yes 14.36%(11,028) 14.1%(1539) 14.41%(9489)
Weather forecast 0.84
Clear 63.04%(48,406) 62.69%(6843) 63.1%(41,563)
Rain 24.76%(19,014) 25.1%(2740) 24.71%(16,274)
Storm 12.19%(9361) 12.2%(1332) 12.19%(8029)
Text message reminder service 20.54 ***
Not yet activated 75.29%(57,806) 76.86%(8389) 75.03%(49,417)
Activated but not sent 16.04%(12,312) 14.61%(1595) 16.27%(10,717)
Sent 8.68%(6663) 8.53%(931) 8.7%(5732)
Factors associated with the examination
No NHS coverage period 0.002
No 90.36%(69,379) 90.34%(9861) 90.36%(59,518)
Yes 9.64%(7402) 9.66%(1054) 9.64%(6348)
Contrast agent 31.36 ***
No 75%(57,583) 77.15%(8421) 74.64%(49,162)
Yes 25%(19,198) 22.85%(2494) 25.36%(16,704)
Price of the examination 6.96 ***
Insured Patients 0±0 0±0 0±0
Other Patients 258.16 ± 117.44 234.99 ± 110.22 261.67 ± 118.11
Waiting list 7.35 ± 7.72 7.92 ± 9.23 7.25 ± 7.44 −7.13 ***
Time allowed 38.43 ± 22.99 38.25 ± 21.29 38.46 ± 23.26 0.35
Descriptive statistics are reported as percentages and number of observations (in parentheses) for categorical variables, as mean ± SD for continuous variables.
The last column reports the t-value or the χ2-value together with the significance symbols

2. The expected times computed sub 1. were cumulated and the extra slots were created and then filled (on a
across booked appointments and, as soon as the result first-come first-served basis) with new incoming
is equal to or greater than the time needed for that reservation requests.
type of examination, an overbooking slot was created
and placed next to the existing booking with the A flowchart in Additional file of the paper (Additional file 1:
highest no-show probability. Figure S1) visually represents this procedure.
3. Such a computational procedure was performed The opportunity created by its actual implementation in
(every evening) 2 days before the day under review the healthcare center allowed us to verify its effectiveness
Parente et al. BMC Health Services Research (2018) 18:185 Page 7 of 12

Table 3 Base Levels for the categorical variables included in the with parameters estimated from real data (μ = − 19.70,
regression model σ = 21.43, ν = 4.35; values in minutes).
Variable Base Level
Gender Male For sake of simplicity, for each examination we
Age group 46–64 assumed a fixed price of €200 and a fixed theoretical
duration of 30′ based on the theoretical duration of the
Insurance Status NHS
10 most frequently asked MRs in the healthcare center.
Type of patient New Patient
Nevertheless, the shorter MR lasts about 15 min, while
Booking confirmation Not confirmed the longest one takes about 50 min. We also assumed
Type of booking Contact Center that, for each day, there were enough new requests from
Day of the week Wednesday patients that could fill the gap generated by no-show.
Month of the year March The overbooking algorithm worked exactly as described
in section An overbooking quasi experiment.
Year 2012
A first set of simulation experiments considered one
Time of the day 8 AM-1 PM
MR scanner and different mean levels of no-show (5, 10,
Long weekend No 15, 30 and 45%) obtained by adjusting the intercept of
Weather forecast Clear our logistic regression model:
Text message reminder service Not yet activated
No NHS coverage period No β0 ¼ β0 þ logðpE =ð1−pE ÞÞ− logðpA =ð1−pA ÞÞ
Contrast agent No

where:
in a quasi-experiment (i.e. we could not randomize over-
booking- and non-overbooking days), comparing the per-  pE is the assumed average probability of no-show
formance relative to days without overbooking. (5%, 10%, 15%, 30% or 45%);
 pA is the overall probability of no-show in the esti-
mated sample.
An overbooking simulation study
We ran a complementary simulation study in which the A second set of simulations used a fixed 15% mean
overbooking strategy was evaluated under different no-show rate, but a different number of active MR scan-
scenarios to get an evidence less influenced by specific ners (1 to 5). For each scenario, 100 replications were
institutional settings and other non-random factors. The performed, performing univariate paired t-tests on the
simulation code is written in R [20] and is available upon equality of the means, with and without overbooking.
request from the corresponding author.
The study simulated the scheduling procedure of the Results
MR examinations considering the same calendar (January The quasi experiment and the simulation study pro-
1 to June 30, 2015) as the overbooking quasi experiment vided complementary evidences: the former regarded
and, for each day, the same number of actual working a non-randomized experiment in a specific real set-
hours (starting 6 AM). For each booked examination, ting; the latter concerned a randomized study under
three random variables were considered: different simulated scenarios.

 the show/no-show behavior (0/1) was simulated A statistical model for show/no-show
through a Bernoulli random variable with The main results of the logistic regression analysis are
probability equal to the no-show probability of summarized in Table 4. Among the variables included
the individual evaluated using his/her values of in the category “Patient’s intrinsic factors”, women
the explanatory variables and the parameters appeared more likely to no-show than men, as well as
estimated as mentioned in section A statistical patients aged 19–45 and over 80 in comparison to
model for show/no-show; patients in the group 46–64. A higher non–attendance
 the actual duration of the examination was generated rate was associated with online reservations and with
through a gamma random variable whose parameters examinations not covered by the NHS, while a lower
were estimated from real data (α = 11.73, β = 0.51, no-show rate was associated to patients providing a
E[X] = 22.83; values in minutes); booking confirmation. Previous history was also strongly
 the patient early/late arrival (scheduled minus arrival predictive: patients with higher rates of previous cancella-
times) was represented by a location-scale Student-T tions or no-shows had higher non–attendance rates, while
Parente et al. BMC Health Services Research (2018) 18:185 Page 8 of 12

Table 4 Logistic regression results Table 4 Logistic regression results (Continued)

Variable Magnetic All wards Variable Magnetic All wards
Resonance (average) Resonance (average)
Intercept −1.684 (−42.67)*** Text message Activated 0.063 (1.84) 0%
reminder service but not sent
Patient’s intrinsic factors
” Sent −0.178 (−4.9) *** 67%
Gender Female 0.113 (7.62) *** 50%
Factors associated
Age group 0–18 −0.003 (− 0.09) 43% with the examination
” 19–45 0.105 (6.2) *** 71% No NHS Yes 0.27 (4.6) *** 86%
” 65–79 −0.118 (−5.39) ***
57% coverage period
” 80+ 0.066 (1.38) 29% Contrast agent Yes 0.018 (0.97) 0%
Insurance status Private 0.371 (4.3) ***
100% Price of the −0.002 (−7.14) *** 50%
examination
Rate of previous 0.198 (5.82) *** 29%
cancellations Waiting list 0.024 (21.79) *** 71%
*** ***
Rate of previous 1.465 (19.97) 71% Time allowed 0.002 (5.7) 43%
no-shows
Type of patient Returned patient − 0.45 (−28.27) *** 100% Goodness of fit measures Magnetic All wards
Resonance (average)
Booking Confirmed −0.98 (−53.38) ***
100%
confirmation R2 0.05 0.19
*** Hosmer-Lemeshow 25.76 15.11
Type of booking Web 0.933 (9.5) 29%
Exogenous factors AIC 59,787.9 16,085.55
Day of the week Monday 0.065 (2.63) ** 14% AUC 0.66 0.77
” Tuesday 0.04 (1.61) 57% Goodness of fit measures – Out of Sample Magnetic All wards
Resonance (average)
” Thursday 0.063 (2.53) * 14%
Hosmer-Lemeshow 38.31 26.52
” Friday 0.008 (0.32) 0%
AUC 0.65 0.75
” Saturday 0.085 (3.11) ** 43%
For the MR ward, estimated parameters, z-value (in parentheses) and significance
” Sunday 0.337 (9.79) *** 50% symbols are reported. For the remaining wards, we reported how many times, in
percentage, each variable is significant at α = 0.05. Significance codes: 0’***’
Month of the year January −0.033 (−0.98) 0%
0.001’**’ 0.01’*’ 0.05
” February 0.052 (1.59) 0%
” April −0.054 (−1.53) 0% those with previous examinations in the healthcare center
” May −0.087 (−2.63) ** 0% had lower propensity to no-show.
” June 0.037 (1.1) 14% When it came to the “Exogenous factors”, all times of
” July 0.001 (0.03) 0%
day were positively associated with higher no-show rates
when compared to the reference 8 AM-1 PM. During the
” August −0.029 (−0.74) 0%
weekend, there was a higher propensity to non–attend-
” September −0.099 (−2.91) ** 0% ance as well as on days in which the weather forecasts pre-
” October 0.067 (1.53) 0% dicted rain. A substantial positive effect came from the
” November −0.22 (−3.35) *** 14% introduction of the text message reminder, which gener-
” December 0.046 (0.63) 0% ated higher attendance rates in patients receiving it.
Year 2013 −0.122 (−6.07) *** 17%
Regarding the “Factors associated with the examin-
ation”, a higher no-show rate was associated with a
” 2014 −0.395 (−12.68) *** 29%
longer waiting list (in days) between the booking and
Time of the day 6 AM–8 AM 0.165 (6.31) *** 100% the examination day, with longer examinations and
” 1 PM–8 PM 0.235 (13.58) *** 57% with appointments performed when the examination
” 8 PM-2 AM (+1d) 0.507 (20.42) *** 100% costs were not covered by the NHS. Higher prices of
Long weekend Yes −0.041 (−1.9) 0% the examination, on the opposite, were related to lower
Weather forecast Rain 0.056 (3.13) ** 29%
no-show rates: we interpreted this result in the sense
that prices are a proxy of the complexity and import-
” Storm 0.027 (1.23) 0%
ance of the service demanded by patients.
As mentioned above, although we detailed results only
for the MR ward, we performed analyses for each ward
(as reported in the Additional file 1) of the healthcare
Parente et al. BMC Health Services Research (2018) 18:185 Page 9 of 12

center. The last column in Table 4 gave the percentage simulation with and without overbooking were statisti-
of how many times each variable was a significant cally significant for all variables considered.
predictor of appointment failure (P ≤ .05) considering all By rising the no-show mean level between 5 and 45%,
wards. This allowed us to identify which variables were the hourly revenue increased significantly (from 2.9% to
more relevant for predicting no-shows in a healthcare 40%), as expected, the waiting time and the overtime
center with heterogeneous wards, each characterized by remained substantially stable, while the idle time re-
its own specificities. The most interesting variables duced considerably.
appeared to be age group, insurance status, type of pa- Table 7 summarize the second set of simulations (vari-
tient (returned or not), booking confirmation and rate of able number of scanners for a mean no-show probability
previous no-shows; on the opposite, the rate of previous of 15%); we assumed that all resources were perfectly
cancellations was not particularly significant across the interchangeable, in the sense that each patient was exam-
wards. About other factors, a significant role was played ined in the first available scanner, not necessarily corre-
by time of the day and text message reminder service. sponding to that indicated in the reservation. The hourly
revenue gain was stable around 13%, as well as the idle
An overbooking quasi experiment time improvement, approximately constant around 3′36″.
Considering Table 5, the overbooking strategy (described The waiting time reduced from 6′42″ to 2′42″, likely as
in section An overbooking quasi experiment) increased an effect of the interchangeability of the resources. Finally,
the hourly revenues by 15.4%, with a statistically signifi- the effect on the overtime was substantially irrelevant, as
cant difference between the means of the hourly revenue indicated by the t-test statistics.
with and without the strategy. The patient’s waiting time
increased by 6′12″; staff’s overtime by 10′, while the idle Discussion
time decreased by 9′06″. Note, however, that the differ- Patient’s non–attendance to booked examinations has
ences between the means of these last three variables in a negative impact on the health system, because indi-
the two subsets were not statistically significant. These viduals who do not show up at booked appointments
results provide evidence that overbooking could con- end up delaying the treatment of other patients, in-
veniently improve the management of the healthcare creasing waiting list, wasting resources and reducing
center, without affecting negatively the quality of the the capability of healthcare providers to invest in med-
care offered to the patient and the costs of the health- ical equipment.
care center. The first contribution of this study is in providing add-
itional evidence on the possible predictors of missing
An overbooking simulation study medical appointments. Using data coming from a
Considering the first set of simulations (Table 6), for a healthcare center located in Southern Italy (with a 14.6%
mean level of no-show of 15%, the hourly revenue in- average level of no-show, consistent with [2, 3]) we built
creased by 12.6% with the introduction of the overbook- a statistical model to predict non–attendance. To this
ing strategy. Under the same conditions, the patient’s aim, we classified each explanatory variable as a
waiting time grew by 3′36″, the overtime increased by “Patient’s intrinsic factor”, an “Exogenous factor” or a
1′30″ (due to longer time needed to complete examina- “Factor associated with the examination”. This classifica-
tions) and the idle time decreased by 4′06″. The t-tests tion represented a first useful contribution of this study
indicate that the differences between the means of the because, for example, the healthcare center may reduce
no-shows working on the factors it can manage (for
example, reminder text messages), since it has little con-
Table 5 Overbooking quasi experiment. (OB=Overbooking) trol on patient’s intrinsic characteristics, factors associ-
Variable Perc. difference Difference t-stats ated with the examination or on the weather.
(with - without OB) (with - without OB)
Regarding the structural characteristics of the patients,
Hourly Revenue 15.4 6.9 2.76 young adults represented the category more likely to miss
(Eur)
scheduled appointments [6, 7, 10]. About the exogenous
Waiting time 3.66 6.2 0.34
(Min)
factors, no-show was related to the time of the day, day of
the week and weather conditions [16]. The patient’s previ-
Idle time (Min) −1.42 −9.1 −1.33
ous history was strongly predictive: individuals with previ-
Overtime (Min) 4.05 10 0.3 ous cancellations or no-shows had higher non–attendance
Number of days 112 rates [2, 7, 21], while those with one or more previous ex-
with OB
aminations in the healthcare center had lower propensity
of days without 62 to no-show [9]. As in previous studies, we found that
OB
non–attendance was associated with the waiting list for an
Parente et al. BMC Health Services Research (2018) 18:185 Page 10 of 12

Table 6 Simulation analysis with one scanner and different levels of no-show
Hourly Revenue (Eur) Waiting time (Min) Idle time (Min) Overtime (Min)
No-show (%) Diff. (%) Average Reject (%) Diff. Average Reject (%) Diff. Average Reject (%) Diff. Average Reject (%)
t-stats t-stats t-stats t-stats
5 2.9 5.76 100 1.2 2.26 56 −0.9 −4.12 100 0.2 0.18 6
10 7.8 11.13 100 2.6 4.85 100 −2.5 −9.19 100 0.2 1.17 7
15 12.6 14.35 100 3.6 6.44 100 −4.1 −12.30 100 0.5 0.44 7
30 26.2 18.87 100 4.9 8.10 100 −9.3 −16.27 100 0.8 0.75 8
45 40.0 20.91 100 4.6 7.33 100 −16.6 −16.97 100 0.9 1.06 18
Reject indicates the percentage of rejections of the t-test at 5%. Difference is computed as with minus without overbooking

appointment (days between booking and the examination) Among exogenous factors, consistently with [4], our
[6, 7]. Examinations booked many days in advance were results suggest that no-show rates can vary according to
more likely to be forgotten or to be not useful anymore time-related factors: we could speculate that, in some
(maybe because, in the meantime, the health issue had hours of the day (i.e. before 6 AM and after 8 PM),
disappeared or, rather, patients required hospitalization). public transportation may be less frequent or even not
We also found a higher no-show rate associated to online available at night or in the weekends.
reservations and to examinations not covered by the NHS A positive effect came from some expedients used to
(or performed during a period not covered by the NHS). reduce missed appointments: a lower propensity to no-
This finding relates to the specific regional healthcare sys- show was found for patients answering to a phone call
tem, but it suggests the more general idea that, when pa- or receiving a text message reminder [5, 8–11]. This
tients had to pay for their examinations, they tended to finding strengthens the existing evidence that “text mes-
miss the appointments more often than when the NHS saging reminders increased attendance at healthcare
paid for the service. We do not have an explanation as of appointments compared to no reminders, or postal
why appointments booked online were more likely to be reminders” [22]. This is an easy strategy to reduce the
missed. We could speculate that patients booking online number of no-shows due to patients’ forgetfulness and it
figure that, if they did not show up for their examination, should be implemented in most outpatient centers.
they could easily book a new appointment without having Web, mobile and, in general, self-service appointment
to talk to a real person, who might have known that the possibilities, in fact, are important technological oppor-
patient had not shown up to a previous appointment. tunities, which may be labor-efficient for the provider
Moreover, the online system used by the healthcare center organization as well as convenient for the patient. The
shows the patient all the available slots, while the call cen- no-show rate in the period without text message
ter’s staff proposes only a few slots to the patient, offering reminder is 14.85% while the non attendance rate in the
more only upon a patient’s request. Of course, the signifi- period with it is 13.80%. As a consequence, if text
cant patient’s intrinsic factors are only under limited con- message reminder was not activated, the overbooking
trol by healthcare providers. Big healthcare providers or strategy would give an even greater contribution in the
payers, such as integrated managed care organizations management of no-show (according to the results of the
or the NHS, may try to implement educational cam- simulation experiment in section An overbooking simu-
paigns to raise the patients’ awareness about the lation study) because there will be an higher non attend-
negative impact of missed appointments. ance rate.

Table 7 Simulation analysis with fixed no-show level at 15% and different number of active MR scanners
Hourly Revenue (Eur) Waiting time (Min) Idle time (Min) Overtime (Min)
MR scanners Diff. (%) Average Reject (%) Diff. Average Reject (%) Diff. Average Reject (%) Diff. Average Reject (%)
t-stats t-stats t-stats t-stats
1 12.4 6.26 100 6.7 7.08 100 −3.3 −12.36 100 0.6 0.79 12
2 13.3 10.32 100 4.4 8.61 100 − 3.5 −18.75 100 0.8 0.47 9
3 13.6 13.30 100 3.5 9.02 100 −3.6 −23.19 100 1.1 0.58 9
4 13.7 15.56 100 3.0 9.66 100 −3.6 −27.34 100 1.3 0.57 12
5 13.8 17.52 100 2.7 9.75 100 −3.6 −30.46 100 2.0 0.78 12
Reject indicates the percentage of rejections of the t-test at 5%. Difference is computed as with minus without overbooking
Parente et al. BMC Health Services Research (2018) 18:185 Page 11 of 12

Among factors associated with the examination, both active MR scanners and rates of no-shows. The results
the price and the time allowed by the healthcare center of the simulations are in line with [4], demonstrating
had a small but significant impact on patient’s no- that overbooking strategies can increase clinic productiv-
show. The tendency to show less often for more ity, reduce resource idle time, although they may in-
expensive examinations may be due to the fact that crease patient’s waiting time and staff’s overtime. An
more expensive procedures may be more important for important policy implication is that the increase of the
the patients. For instance, price may be a proxy for the number of active MR scanners, for the same average
importance of the examination because doctors may level of no-shows, may improve the efficiency of the
tend to prescribe MRs only if in the lack of alternative healthcare center.
cheaper diagnostics. As of the positive impact of longer Regarding the limitations, our analysis was performed
MR procedures, it may be related to the higher anticipated on just one healthcare center and, accordingly, it is not
stress, the longer the time spent in a MR scanner. necessarily portable to others. Moreover, such a healthcare
Our predictive model (independent variables, magni- center operates within the Italian NHS, whose budget
tude and significance of the corresponding coefficients) covers most medical examinations. This underlines the
cannot be piece-by-piece transposed to different socio- importance of estimating model parameters on variables
economical contexts and health systems. However, we and data reflecting the specific organizational options in
show that is possible to reliably predict no-shows using each healthcare center. A second limitation when consid-
context-specific patient’s intrinsic factors, factors associ- ering several active MR scanners: we assumed that the
ated with the examination and exogenous factors. patient was diagnosed with the first resource available,
Prediction models built with local data can then be used while, in other settings, one encounters technical limits to
to develop context-specific overbooking algorithms. substitutability across scanners (and hence a need to stick
Another contribution of this study is the evaluation of to the one listed in the reservation). This case was not ex-
the overbooking strategy built on our no-shows predic- plored in our simulation study.
tors. When not able to intervene on non-attendance The overbooking procedure is dictated by an
causes, the center may resort to overbooking to manage organizational choice by the center: more frequent
its effects. It represents a trade-off between the pro- updates may strain the management of the high incom-
vider’s productivity and the waiting list for a new ing phone call flow to book an examination. Further
appointment on the one hand, and overcrowding and refinements could come from the availability of patients’
delays due to patients showing up in larger numbers socioeconomic conditions, levels of education, distance
than expected at a given time of day [4], on the other. to reach the healthcare center, which could be significant
We contribute to the previous literature on overbooking for the prediction of the no-show [2, 5–7]. Moreover,
in healthcare [4, 14–16] by testing in a real setting a clinical information about the patients could signal the
strategy, which used a fairly complex predictive model. urgency of the examination and/or the possibility of
We showed a significant productivity gain (+ 15.4%) in selecting homogeneous patient cohorts.
terms of hourly revenue, increasing delays suffered by
patients only by non-significant amounts. This suggests
that overbooking could improve healthcare centers and, Conclusions
indirectly, the quality of service. This study contributed to the literature on appointment
We speculate that the negligible effects of overbooking attendance and overbooking in healthcare by showing,
on patients were due to: in a setting with real data, that overbooking may imply
benefits with limited side effects.
 the quality of the underlying predictive model, Moreover, we strengthened the evidence of a predict-
which took into account patient’s intrinsic factors, ive model testing most of the variables considered by
exogenous factors and other characteristics previous studies in a single model, using a large dataset
associated with the examination and was based on in a specific socio-economic context. This predictive
a considerable amount of observations; model also introduced two new variables influencing no-
 the fact that the predictive model was built on show: online booking and insurance/NHS coverage. This
context-specific data; study provided an encouraging evidence that overbook-
 the size of this healthcare center operations, with ing procedures could improve the management of
five MR scanners working alongside one another. healthcare centers. The current upward trend regarding
the availability of data and the access to complex data
In order to complement these findings with a more analysis tools, allows an increasing number of healthcare
general evidence, we also proposed a simulation study organizations to adopt overbooking practices based on
considering various scenarios across several numbers of well performing predictive models, such as that derived
Parente et al. BMC Health Services Research (2018) 18:185 Page 12 of 12

in this paper. The quasi experiment gave real setting- Received: 6 March 2017 Accepted: 1 March 2018
evidence that overbooking could improve efficiency -
and, indirectly, the quality of services - while increasing References
waiting time suffered by patients only by small and 1. Rust CT, Gallups NH, Clark WS, Jones DS, Wilcox WD. Patient appointment
statistically non-significant amounts. This evidence en- failures in pediatric resident continuity clinics. Arch Pediatr Adolesc Med.
1995;149:693–5.
courages further studies to strengthen the evidence 2. Collins J, Santamaria N, Clayton L. Why outpatients fail to attend their
about an extended adoption of overbooking strategies in scheduled appointments: a prospective comparison of differences
healthcare centers. between attenders and non-attenders. Aust Health Rev. 2003;26:52–63.
3. Moore CG, Wilson-Witherspoon P, Probst JC. Time and money: effects of
no-shows at a family practice residency clinic. Fam Med. 2001;33:522–7.
4. Laganga LR, Lawrence SR. Clinic overbooking to improve patient access and
Additional file increase provider productivity. Decis Sci. 2007;38:251–76.
5. Campbell JD, Chez RA, Queen T, Barcelo A, Patron E, et al. J Women’s
Additional file 1: Web appendix with the logistic regression results Health Gender Based Med. 2000;9:891–5.
for the other wards (Echography, Echo Doppler, Mammography, 6. Dreiher J, Froimovici M, Bibi Y, Vardy DA, Cicurel A, Cohen AD.
Orthopantomography, Radiography, Computer-Assisted Tomography) Nonattendance in obstetrics and gynecology patients. Gynecol Obstet
and the description and the representation of the flow chart. Investig. 2008;66:40–3.
(PDF 668 kb) 7. Lee VJ, Earnest A, Chen MI, Krishnan B. Predictors of failed attendances in a
multi-specialty outpatient Centre using electronic databases. BMC Health
Serv Res. 2005;5–51.
Abbreviations 8. Guse CE, Richardson L, Carle M, Schmidt K. The effect of exit-interview
MR: Magnetic resonance; NHS: National health service; OB: Overbooking patient education on no-show rates at a family practice residency clinic.
J Am Board Fam Pract. 1997;16:399–404.
9. Johnson BJ, Mold JW, Pontious JM. Reduction and management of
Acknowledgements
no-shows by family medicine residency practice exemplars. Ann Fam Med.
The views expressed in the paper are those of the authors and do not
2007;5:534–9.
involve the responsibility of the Corte dei conti.
10. Lasser KE, Mintzer IL, Lambert A, Cabral H, Bor DH. Missed appointment
rates in primary care: the importance of site of care. J Healthcare Poor
Funding Underserved. 2005;16:475–86.
Not Applicable. 11. Perron NJ, Dao MD, Kossovsky MP, Miserez V, Chuard C, Calmy A,
Gaspoz J-M. Reduction of missed appointments at an urban primary
care clinic: a randomised controlled study. BMC Fam Pract.
Availability of data and materials
2010;11:79.
The data were not available because this healthcare center preferred not to
12. Kaandorp GC, Koole G. Optimal outpatient appointment scheduling.
make public data that may reveal corporate policies other than those object
Healthcare Manage Sci. 2007;10:217–29.
of this study. Nevertheless, the code used to analyze the data, written in R, is
13. Chakraborty S, Muthuraman K, Lawley M. Sequential clinical scheduling
available upon request from the corresponding author.
with patient no-shows and general service time distributions. IIE Trans.
2010;42:354–66.
Authors’ contributions 14. Muthuraman K, Lawley M. A stochastic overbooking model for
CAP, DS, FC, GMG were involved in the conception and design of the study, outpatient clinical scheduling with no-shows. IIE Trans. 2008;40:820–37.
the interpretation of the data and of the results and agree to be 15. Zeng B, Turkcan A, Lin J, Lawley M. Clinic scheduling models with
accountable for all aspects of the work. CAP performed the analysis. All the overbooking for patients with heterogeneous no-show probabilities.
authors have read, discussed and approved the results and contributed to Ann Oper Res. 2010;178:121–44.
the final manuscript. 16. Laganga LR. Lean service operations: reflections and new directions for
capacity expansion in outpatient clinics. J Oper Manag. 2011;29:422–33.
17. Ferré F, De Belvis AG, Valerio L, Longhi S, Lazzari A, Fattore G,
Ethics approval and consent to participate
Ricciardi W, Maresso A. Italy: health system review. Health Syst Transit.
This study used retrospective and anonymized administrative data of
2014;16:1–168.
patients who gave their consent to the use. Therefore this type of study was
18. Lee CH, Dershaw DD, Kopans D, Evans P, Monsees B, Monticciolo D,
exempted from approval by an ethics committee.
Brenner RJ, Bassett L, Berg W, Feig S, Hendrick E, Mendelson E, D'Orsi
C, Sickles E, Burhenne LW. Breast cancer screening with imaging:
Consent for publication recommendations from the Society of Breast Imaging and the ACR
Not applicable. on the use of mammography, breast MRI, breast ultrasound, and
other technologies for the detection of clinically occult breast cancer.
J Am Coll Radiol. 2010;7:18–27.
Competing interests
19. Hosmer Jr, David W, Lemeshow S, Rodney X. Sturdivant. Applied logistic
The authors declare that they have no competing interests.
regression. Ed. 2. Wiley; 2004.
20. R Core Team. R: a language and environment for statistical
computing. R foundation for statistical computing. Vienna: R
Publisher’s Note Foundation for Statistical Computing; 2015. http://www.R-project.org/
Springer Nature remains neutral with regard to jurisdictional claims in 21. Satiani B, Miller S, Patel D. No-show rates in the vascular laboratory: analysis
published maps and institutional affiliations. and possible solutions. J Vasc Interv Radiol. 2009;20:87–91.
22. Car J, Gurol-Urganci I, de Jongh T, Vodopivec-Jamsek V, Atun R. Mobile
Author details phone messaging reminders for attendance at healthcare appointments.
1
IRCCS SDN, Napoli, Via E. Gianturco 113, 80143 Naples, Italy. 2Department of Cochrane Database of Systematic Reviews 2012, Issue 7. Art. No.: CD007458.
Accounting, Management and Economics, University of Naples Parthenope, https://doi.org/10.1002/14651858.CD007458.pub2.
Via Generale Parisi, 13, 80132 Naples, Italy. 3Corte dei conti, Sezione regionale
di controllo per la Lombardia, via Marina 5, 20121 Milan, Italy. 4Department
of Statistics, Informatics and Applications (DiSIA) G. Parenti, University of
Florence, Viale Giovanni Battista Morgagni, 59, 50134 Florence, Italy.
BioMed Central publishes under the Creative Commons Attribution License (CCAL). Under
the CCAL, authors retain copyright to the article but users are allowed to download, reprint,
distribute and /or copy articles in BioMed Central journals, as long as the original work is
properly cited.