Estimation of sickness absenteeism among Italian healthcare workers during seasonal influenza epidemics
Estimation of sickness absenteeism among Italian healthcare workers during seasonal influenza epidemics
Maria Michela Gianino 0 1
Gianfranco Politano 1
Antonio Scarmozzino 1
Lorena Charrier 0 1
Marco Testa 0 1
Sebastian Giacomelli 0 1
Alfredo Benso 1
Carla Maria Zotti 0 1
0 Department of Public Health Sciences and Pediatrics, Università di Torino , Torino , Italy , 2 Department of Control and Computer Engineering , Politecnico di Torino, Torino, Italy, 3 AOU Città della salute e della Scienza, Torino , Italy
1 Editor: Benjamin J. Cowling, University of Hong Kong , HONG KONG
Retrospective observational study.
To analyze absenteeism among healthcare workers (HCWs) at a large Italian hospital and
to estimate the increase in absenteeism that occurred during seasonal flu periods.
The absenteeism data were divided into three ªepidemic periods,º starting at week 42 of one
year and terminating at week 17 of the following year (2010±2011, 2011±2012, 2012±
2013), and three ªnon-epidemic periods,º defined as week 18 to week 41 and used as
baseline data. The excess of the absenteeism occurring among HCWs during periods of
epidemic influenza in comparison with baseline was estimated. All data, obtained from
Hospital's databases, were collected for each of the following six job categories: medical
doctors, technical executives (i.e., pharmacists), nurses and allied health professionals (i.e.,
radiographers), other executives (i.e., engineers), nonmedical support staff, and
administrative staff. The HCWs were classified by: in and no-contact; vaccinated and unvaccinated.
5,544, 5,369, and 5,291 workers in three years were studied. The average duration of
absenteeism during the epidemic periods increased among all employees by +2.07 days/
person (from 2.99 to 5.06), and the relative increase ranged from 64±94% among the
different job categories. Workers not in contact with patients experienced a slightly greater
increase in absenteeism (+2.28 days/person, from 2.73 to 5.01) than did employees in
contact with patients (+2.04, from 3.04 to 5.08). The vaccination rate among HCWs was below
3%, however the higher excess of absenteeism rate among unvaccinated in comparison
with vaccinated workers was observed during the epidemic periods (2.09 vs 1.45 days/
The influenza-related absenteeism during epidemic periods was quantified as totaling more
than 11,000 days/year at the Italian hospital studied. This result confirms the economic
impact of sick leave on healthcare systems and stresses on the necessity of encouraging
HCWs to be immunized against influenza.
The annual occurrence of seasonal flu epidemics and subsequent work absenteeism, coupled
with the low immunization coverage achieved among healthcare workers (HCWs), may have a
significant impact on patient health, requiring targeted policy interventions.
The WHO has estimated that as a result of seasonal influenza epidemics, 5±15% of the
population is affected by upper respiratory infections, and 3±5 million cases of severe illness and
between 250,000 and 500,000 deaths occur each year worldwide .
Globally, access to vaccination is considered insufficient in many populations, including
high-risk groups. Moreover, the WHO objective of achieving vaccination coverage of at least
50% by 2006 and 75% by 2010 in the elderly population and among at-risk individuals was not
A large study conducted in Europe [
] reported vaccination rates in the general population
ranging from 10±30%, with the lowest rate of vaccination identified in those under 50 years of
age. Additionally, in elderly subjects ( 65 years), vaccination rates ranged from a minimum of
14% (Ireland) to a maximum of 70% (UK). Thus, immunization rates were low, even among
patients with chronic respiratory or cardiovascular diseases (25±60%) and the elderly (17±
Low immunization among HCWs is a major issue both because of the risk of transmitting
vaccine-preventable infections to patients, and especially those at high risk, and given the need
to maintain high health personnel availability during epidemics. In healthcare settings,
influenza, which is spread by droplet transmission, may be introduced by visitors, patients, and
staff, with serious consequences for elderly and immunocompromised patients, and patient
isolation may be insufficient to contain influenza transmission [
]. Available data demonstrate
that despite 30 years of official recommendations, the immunization rate among HCWs in
Europe rarely exceeds 30±40% [
In Italy, between 2000 and 2015, vaccination coverage in the general population ranged
from 13±19%. The coverage of the elderly population (the first target of vaccination) exceeded
65% only a few times between 1999 and 2015, while among HCWs, the data regarding
coverage during the pandemic period (2009±2010) suggested vaccination coverage of only 15%.
Although there has been some general interest in analyzing the correlation between work
absenteeism (in the general population as well as among HCWs) and influenza epidemics,
obtaining the data necessary to allow for precise quantification is difficult; this difficulty
mainly derives from challenges in obtaining comparable data due to different policies for
recording work absenteeism in different countries, leading to different levels of sensitivity and
2 / 12
After the 2009 influenza pandemic, several publications compared work absenteeism
related to the pandemic with absenteeism during periods of seasonal epidemics. One study of
absenteeism among HCWs in Hong Kong [
] during a seasonal epidemic highlighted 8.4%
and 26.5% excesses in absenteeism due to any cause and respiratory diseases, respectively.
Moreover, during the study period (2004±2009), the average durations of an absence from
work due to illness overall and respiratory illness in particular amounted to 2.3 days and 1.39
days, respectively. Meanwhile, a Canadian study evaluating the general population estimated a
12% increase in absenteeism per year due to the seasonal flu, with an average loss of 14
working hours per worker [
]. Another study conducted on workers aged over 50 years in the USA
reported an average loss of 1.3 working days due to influenza-like illness (ILI) [
]. Finally, a
study conducted in the UK [
] estimated a 10±12% increase in HCW absenteeism due to
influenza or ILI.
A systematic review published in 2014 highlighted the effectiveness of HCW vaccination
against influenza in significantly reducing mortality from all causes and ILI [
only a few studies of good quality and at low risk of bias have determined vaccination's
preventive efficacy and its effective impact on absenteeism [10±17]. Additionally, the economic
impact that a preventive vaccination campaign could have on reducing work absenteeism is a
topic of interest for both researchers and national healthcare systems.
The aims of the present study were to analyze absenteeism among HCWs at a large Italian
hospital and to estimate the increase in absenteeism that occurred during seasonal flu periods.
Materials and methods
The AOU ªCittà della salute e della Scienzaº in Turin is a complex of four interconnected
hospitals (Molinette, OIRM, S. Anna, and CTO) with more than 1,700 beds and is the main
teaching hospital of the University of Turin's School of Medicine. We conducted this study based
on data from Molinette, which has approximately 5,500 workers, accounting for
approximately 45% of the center's employees.
In this study, we analyzed data from the three consecutive years following the influenza
pandemic of 2009, during which seasonal influenza outbreaks were of medium intensity.
Absenteeism data were obtained from the hospital's Personal Unit Database and included
the number of days of paid sick leave during the periods of July 2010 to June 2011, July 2011 to
June 2012, and July 2012 to June 2013. Such database also comprised for every employee a set
of attributes like as personal data, job category, position, work place.
From the Occupational Health Unit of Molinette Hospital we obtained data on influenza
vaccination for each employee and we merged this database with the Personal Unit Database
in order to obtain a single, comprehensive database.
We focused on ªsporadic absences,º defined as unplanned sickness absenteeism due to any
We could not obtain a dataset including only ILI-related and acute respiratory infection
(ARI)-related absences because based on the Italian policy regarding absenteeism records in
the workplace, it is not compulsory to note the medical diagnosis reported on the sickness
absence certificate issued by the medical practitioner.
We divided the data from the different years into three ªepidemic periods,º starting at week
42 of one year and terminating at week 17 of the following year (2010±2011, 2011±2012, and
2012±2013), and three ªnon-epidemic periods,º which were defined as week 18 to week 41 and
used as baseline data. The epidemic period counted 196 days and the non-epidemic period
counted 168 days. During these three consecutive years that followed the influenza pandemic
3 / 12
of 2009, outbreaks of moderate intensity occurred with the following ILI epidemic incidence
· 2010±2011: 103/1,000 person-years;
· 2011±2012: 86/1,000 person-years;
· 2012±2013: 105/1,000 person-years.
Data for Italian influenza epidemics were obtained from Influnet, the Italian sentinel
influenza surveillance network. Influnet specifically comprises organized networks of primary care
physicians, and mostly general practitioners (GPs), covering at least 1±5% of the population.
Between week 42 and week 17, sentinel physicians reported the weekly number of patients
with ILI, ARI, or both to the national center for influenza surveillance [
In this study, we also compared ILI morbidity data for the three evaluated seasonal
epidemic periods (provided by the regional epidemiological service (SEReMI)) with absenteeism
rates at the target hospital during the same periods.
Individual sickness absenteeism data were grouped for each of the following job categories:
1. medical doctors;
2. technical executives (i.e., pharmacists, dieticians, and chemists);
3. nurses and allied health professionals (i.e., radiographers, therapists, and laboratory
4. other executives (i.e., engineers, lawyers, analysts, and statistical and administrative staff);
5. nonmedical support staff (i.e., ward assistants and cleaning staff);
6. administrative staff.
The overall personnel were also grouped into two categories (in-contact and no-contact),
depending on the nature of their work relationship with patients. The ªin-contactº category
included all workers who were engaged in direct contact with patients during admission,
diagnosis, treatment, and/or follow-up. The ªno-contactº category included all workers who did
not work in proximity to patients and who therefore had a lower risk of transmitting disease.
The study protocol was approved by the Directorate-General of AOU (November 26,
We developed a general and robust R [
] pipeline that was able to process the input data
through the following steps:
· input preparation;
· descriptive analysis;
· comparison with reference data;
· stratification and risk analysis;
Input preparation. One set of csv files was input into the pipeline per solar year. In these
files, each row represented an employee, and each column described one of several attributes
4 / 12
used for stratification (e.g., position, contact with patients, sex) and the number of days of
absence during each of the 52 weeks of the year. The data were also pre-processed by merging
two files at a time and extracting representative data for a given flu-year (i.e., the flu-year starts
from week 27 of a given year and lasts until week 26 of the next year). This transformation
allowed us to better assess the dynamics of a single flu epidemic whose peak occurred during
the winter and spanned across years.
Descriptive analysis. The pre-processed data were then descriptively analyzed. For each
stratification variable, we computed its frequency distribution within the available classes.
Afterward, the pipeline produced graphical output to visually describe the trends in the flu
absenteeism occurring over the years analyzed. To better interpret the data, the trends were
interpolated by fitting a lowess curve that smoothed the intra-week variance due to seasonal
vacations and other possible sources of absenteeism (i.e., parental care and other reasons for
Comparison with reference data. To increase data comparability between different
populations that were most likely heterogeneous in terms of distribution, the results were
represented in terms of days lost per person due to illness.
If reference data were available (i.e., regional/government observational flu data), the
pipeline was able to graphically produce a comparison between the experimental and reference
data. In our case, the reference data showed a cleaner signal, mainly because of the high
specificity of these data (only representing certified ILI or ARI cases). In contrast, the experimental
data exhibited a higher level of noise, which resulted in a greater basal offset and, consequently,
a more compressed signal.
Stratification and risk analysis. After the descriptive analysis, the pipeline employed the
] and meta [
] packages to compute a risk analysis score for each flu year and a
cumulative meta-analysis score to determine the average intra-year risk of absenteeism due to the
Risk analysis (RA) is a widely addressed method in epidemiology to identify and quantify,
in terms of numerical probability, the relationship (risk) between the exposure to a given
condition (flu period/non-flu period) and a given effect (absenteeism /no absenteeism).
In the proposed study, individuals in the flu period (the "case" group) are compared with
individuals in the non-flu period (the "control" group). We constructed a 2×2 confusion
matrix (Table 1) representing the count of absenteeism days of cases (A), absenteeism days of
controls (B), no absenteeism days of cases (C) and no absenteeism days of controls (D).
From the confusion matrix we computed the cumulative incidence (CI), which is an
estimate of the risk of absenteeism for each exposure group and is computed as follow:
CIexposed A=A C and CIunexposed B=B D:
Furthermore, in order to retain a more robust overall insight into absenteeism risk over the
three-year timespan, we applied a meta-analysis approach by correcting with fixed-effect
a The duration of the flu and non-flu periods is counted on the basis of epidemiologic and virological surveillance by general practitioners
5 / 12
estimate, using the Mantel-Haenszel method [
], to properly manage differences in
population across the three years timespan.
Output. In order to elucidate the difference in days of a condition between an exposed
and an unexposed population we resorted to the following mathematical expression:
non epidemic days
The Excess absenteeism score conservatively quantified the flu effect in terms of the excess
of absenteeism days observed during the flu periods and for years/person.
The numbers of HCWs employed at the target hospital were 5,544, 5,369, and 5,291 during the
three years under study (2010, 2011, and 2012, respectively; 73% female). The reduction
observed in the number of workers was predominantly caused by retirement, no data were
In the three years under analysis, there were no significant differences in the distribution of
workers by job category, age class, percentage of employees working in direct contact with
patients, or vaccination rate. Most employees were nurses and allied health professionals, were
aged between 40 and 59 years, and worked in direct contact with patients, and only 2.2±2.9%
of employees were vaccinated (Table 2).
Fig 1 shows that the sickness absenteeism rates in the Piemonte region increased during
each of the epidemic periods, and there were discernible peaks between the 51st week and the
11th week in 2010±2011 and between the 52nd week and the 14th week in both 2011±2012 and
2012±2013. Further supporting the hypothesis that the peaks in absenteeism were related to
the influenza epidemic, Fig 2 shows that during the same period, a similar trend was observed
in ILI morbidity in the local community within the Piemonte region.
Given that the vaccination rate among HCWs was below 3%, the overall trend in
absenteeism was not significantly affected by the inclusion or exclusion of vaccinated workers.
* Medical doctors: i.e., physicians and radiologists, Technical executives: i.e., pharmacists, dieticians, biologists, chemists, and similar professions, Nurses
and allied health professionals, i.e., radiographers, therapists, and laboratory technicians, Other executives: i.e., engineers, lawyers, analysts, and
statistical and administrative staff, Nonmedical support staff: i.e., ward assistants and cleaning staff
6 / 12
Fig 1. Weekly sickness absenteeism rates among HCWs.
The average duration of absenteeism during the epidemic period increased among all
employees by +2.07 days/person (from 2.99 to 5.06 days/person).
Table 3 shows the increase in sick leave during the epidemics among employees in each job
category, with the relative increase ranging from 64±94% among the different types of staff.
In comparison with other job categories, the absolute increases in absenteeism were highest
among nonmedical support staff (+3.4), administrative staff (+2.15) and nurses and allied
health professionals (+1.95). These categories also had higher levels of absenteeism during
non-epidemic periods (5.17, 3.07, 2.75 days, respectively) in comparison with other categories.
The ranking of the absenteeism rates by job category during non-epidemic and epidemic
periods were comparable.
Workers not in contact with patients experienced a slightly greater increase in absenteeism
than did employees in contact with patients. The absolute value of the observed excess
absenteeism was 2.28 days/person, which was equivalent to an excess of approximately 84% in
relative terms. Workers not in contact with patients exhibited lower rates of absenteeism during
non-epidemic periods in comparison with workers in contact with patients (2.73 vs 3.04 days/
7 / 12
Fig 2. Morbidity rates associated with influenza epidemics in the Piemonte region.
The absenteeism rate among workers without vaccination during the epidemic periods was
approximately 1.5 times higher than the rate observed among vaccinated employees. The
absolute and relative increases were 2.01 days/person and approximately 70%, respectively.
Our study showed that there was an increase in absenteeism among hospital workers during
periods of epidemic influenza in Italy (average of +2.07 days/person). Compared with the
average of absenteeism during non-epidemic periods, used as baseline data, this absolute increase
correlated with a relative increase of 70% (from 2.99 to 5.06 days/person).
This finding is in agreement with the results of previous studies. In 1980±1981, an epidemic
of influenza A in Winnipeg resulted in a nearly 2-fold increase in work time lost by hospital
workers (nurses and support personnel) compared with data for the remainder of the year
excluding the epidemic period [
]. In Canada, another study demonstrated that there was a
significant difference between the absenteeism rates among employees in high-risk
departments during the 1987±1988 influenza season and the non-influenza season during the same
year, with an approximately 35% higher absenteeism rate observed during the influenza season
In contrast, a study in Hong Kong based on data collected over a six-year period found a
modest increase in absenteeism among HCWs during influenza epidemics (+8.4%). Finally,
during two epidemics (1993±1994 and 1996±1997), another study conducted in the UK
detected only a limited change in the rates of sickness absenteeism [
These discrepant findings could be explained by differences in the intensity, frequency, and
duration of the epidemics; the strains of influenza virus involved in the epidemics; or the
methodology employed in the studies, such as the study period evaluated, the staff type analyzed
and the size of the sample population of workers, all of which may impact the sickness
absenteeism estimates. Nevertheless, our results, which were supported by morbidity trends in the
Piemonte region, very likely represent the typical impact of seasonal influenza in Italy.
The rate of absenteeism observed in this study increased significantly within all job
categories during epidemic periods. An important factor to consider when quantifying the impact of
influenza epidemics on HCWs is the relationship between employment role (i.e., medical
doctors, technical and other executives, nurses and allied health professionals, nonmedical support
staff, and administrative staff) and the length of the sick leave. In our study, employees
belonging to the first three job categories returned to work, on average, after less than two days of
sick leave during epidemic periods, whereas the workers in the other categories had an average
sick leave duration of approximately five days. The comparison between the epidemic and
non-epidemic periods showed that the increase in absenteeism during epidemic periods was
lower among workers in the first three categories (on average less than 1 day/person) than
among workers in the other three categories (on average more than 2 days/person); this
finding may be attributable to the fact that the medical and executive personnel (approximately
17% of personnel) may recognize that the hospital cannot obtain coverage for their positions
when they take sick leave and therefore may feel obligated to return to work as early as
We hypothesized that due to the low vaccination coverage that characterizes Italian HCWs
(such as those included in the study), vaccination has not had a substantial impact on work
absenteeism trends, and our results confirm this hypothesis. Nevertheless, our study showed
that unvaccinated employees used approximately 3.2 days of additional sick leave per person
during the influenza season compared with vaccinated employees. Interestingly, the difference
in absenteeism between vaccinated and unvaccinated HCWs was also evident also during the
non-epidemic periods, with unvaccinated HCWs having, on average, 2.5 additional
days/person lost due to sick leave. This discrepancy may be explained by the possibility that vaccines
also protect employees from illness during non-epidemic periods. Moreover, the greater
difference in absenteeism observed during the epidemic periods supports the beneficial effect of
influenza vaccination on reducing absenteeism, as observed in previous studies [
These results could be useful to populate mathematical models that allow to estimate the
effects of vaccination interventions; the experiences of models developed [
] are useful for
describing the changes in the dynamics of influenza in relation to the behavioral changes of
the affected population (vaccinations, preventive measures). Stationary pattern or time-space
9 / 12
transitions can also be described in predictive mathematical models and absenteeism data can
help to read both behavioral and economic changes.
In the present study, the rates of absenteeism were slightly higher during both epidemic
and non-epidemic periods among workers in direct contact with patients. However, during
the epidemic periods, there was a slightly lesser increase in the use of sick leave among workers
having direct contact with patients relative to those with no patient contact. Although
influenza is a community infection, so contact with patients may not be its main mode of
transmission, the results showed that HCWs with direct patient contact were at increased risk of
becoming infected and experiencing absenteeism year-round.
This study has some limitations. First, the use of the data referring to sporadic absences,
defined as unplanned sick leaves due to any cause, and no data referring only ILI-related and
acute respiratory infection (ARI)-related absences, might be a limitation. The performance of
similar analyses using GP-certified sick leave data may lead to more robust estimates. However
the results reported in this work are statistically valid and the hypothesis that the peaks in
absenteeism were related to the influenza epidemic, is supported by the fact that during the
same period, a similar trend was observed in ILI morbidity in the local community within the
Second, our study analyzed data from a database, and not a prospective cohort however we
utilized data collected over 3 years, and our study sample covered a very large population of
HCWs spanning all job categories.
In conclusion, at Molinette Hospital, considering the average of +2.07 days/person and all
employees, influenza-related absenteeism during the considered epidemic periods was
quantified as totaling more than 11,000 days/yearº. This result confirms the economic impact of
influenza-related absenteeism on healthcare systems. The difference in sickness absenteeism
between vaccinated and unvaccinated HCWs stresses the necessity of encouraging HCWs to
be immunized against influenza.
We thank Doctor Carlo Conte for his efforts to provide absenteeism data, Occupational Health
Unit Executive Doctor Fabrizio Meliga for data on influenza vaccination coverage at the
hospital, and Doctor Donatella Tiberti (SEReMI) for ILI morbidity data for the Piemonte region.
Financial support: None reported.
Potential conflicts of interest: All authors report no conflicts of interest relevant to this
Conceptualization: Maria Michela Gianino, Carla Maria Zotti.
Data curation: Antonio Scarmozzino.
Formal analysis: Gianfranco Politano, Alfredo Benso.
Investigation: Lorena Charrier, Marco Testa, Sebastian Giacomelli.
Methodology: Maria Michela Gianino, Gianfranco Politano, Alfredo Benso, Carla Maria
Supervision: Maria Michela Gianino, Carla Maria Zotti.
Writing ± original draft: Maria Michela Gianino, Gianfranco Politano, Alfredo Benso.
10 / 12
Writing ± review & editing: Maria Michela Gianino, Gianfranco Politano, Antonio
Scarmozzino, Alfredo Benso, Carla Maria Zotti.
Weingarten S, Staniloff H, Ault M, Miles P, Bamberger M, Meyer RD. Do hospital employees benefit
from the influenza vaccine? A placebo-controlled clinical trial. J Gen Intern Med. 1988; 3(1): 32±7.
11 / 12
2. World Health Organization (WHO) website influenza fact sheet [Internet] . WHO. 2005 [cited 2016 Nov 14 ]. http://www.who.int/mediacentre/factsheets/en/
World Health Organization 56th World Health Assembly. Prevention and control of influenza pandemics and annual epidemics [Internet] . 2003 [cited 2016 Nov 14 ]. http://www.who.int/immunization/sage/1_ WHA56_19_ Prevention_and_control_of_influenza_pandemics .pdf
3. Blank PR , Schwenkglenks M , Szucs TD . Vaccination coverage rates in eleven European countries during two consecutive influenza seasons . J Infect . 2009 ; 58 ( 6 ): 446 ± 58 . https://doi.org/10.1016/j.jinf. 2009 . 04 .001 PMID: 19446340
4. Bridges CB , Kuehnert MJ , Hall CB . Transmission of influenza: implications for control in health care settings . Clin Infect Dis Off Publ Infect Dis Soc Am . 2003 ; 37 ( 8 ): 1094 ± 101 .
5. Ip DKM , Lau EHY , Tam YH , So HC , Cowling BJ , Kwok HKH . Increases in absenteeism among health care workers in Hong Kong during influenza epidemics , 2004 ± 2009 . BMC Infect Dis . 2015 ; 15 : 586 . https://doi.org/10.1186/s12879-015-1316-y PMID: 26715075
6. Schanzer DL , Zheng H , Gilmore J . Statistical estimates of absenteeism attributable to seasonal and pandemic influenza from the Canadian Labour Force Survey . BMC Infect Dis . 2011 ; 11 : 90 . https://doi. org/10.1186/ 1471 -2334-11-90 PMID: 21486453
7. Akazawa M , Sindelar JL , Paltiel AD . Economic costs of influenza-related work absenteeism . Value Health J Int Soc Pharmacoeconomics Outcomes Res . 2003 ; 6 ( 2 ): 107 ± 15 .
8. Keech M , Scott AJ , Ryan PJ . The impact of influenza and influenza-like illness on productivity and healthcare resource utilization in a working population . Occup Med Oxf Engl . 1998 ; 48 ( 2 ): 85 ± 90 .
9. Ahmed F , Lindley MC , Allred N , Weinbaum CM , Grohskopf L . Effect of influenza vaccination of healthcare personnel on morbidity and mortality among patients: systematic review and grading of evidence . Clin Infect Dis Off Publ Infect Dis Soc Am . 2014 ; 58 ( 1 ): 50 ± 7 .
11. Nichol KL , Lind A , Margolis KL , Murdoch M , McFadden R , Hauge M , et al. The effectiveness of vaccination against influenza in healthy, working adults . N Engl J Med . 1995 ; 333 ( 14 ): 889 ± 93 . https://doi.org/ 10.1056/NEJM199510053331401 PMID: 7666874
12. Olsen GW , Burris JM , Burlew MM , Steinberg ME , Patz NV , Stoltzfus JA , et al. Absenteeism among employees who participated in a workplace influenza immunization program . J Occup Environ Med . 1998 ; 40 ( 4 ): 311 ± 6 . PMID: 9571521
13. O 'Reilly FW , Stevens AB . Sickness absence due to influenza . Occup Med Oxf Engl . 2002 ; 52 ( 5 ): 265 ± 9 .
14. Tuccillo E , Corona G , Franco G. [ Evaluation of the efficacy of an influenza vaccine campaign among health personnel for the prevention of influenza-like illness (ILI)] . G Ital Med Lav Ergon . 2003 ; 25 Suppl (3 ): 155 ± 6 .
15. Postma MJ , Jansema P , Scheijbeler HWKFH , van Genugten MLL. Scenarios on costs and savings of influenza treatment and prevention for Dutch healthy working adults . Vaccine . 2005 ; 23 ( 46 ±47): 5365 ± 71 . https://doi.org/10.1016/j.vaccine. 2005 . 06 .007 PMID: 16029916
16. Mota NV y VP, Lobo RD , Toscano CM , Pedroso de Lima AC , Souza Dias MB , Komagata H , et al. Costeffectiveness of sick leave policies for health care workers with influenza-like illness , Brazil , 2009 . Emerg Infect Dis . 2011 ; 17 ( 8 ): 1421 ±9. https://doi.org/10.3201/eid1708.101546 PMID: 21801619
17. Van Buynder PG , Konrad S , Kersteins F , Preston E , Brown PD , Keen D , et al. Healthcare worker influenza immunization vaccinate or mask policy: strategies for cost effective implementation and subsequent reductions in staff absenteeism due to illness . Vaccine . 2015 ; 33 ( 13 ): 1625 ±8. https://doi.org/10. 1016/j.vaccine. 2015 . 01 .048 PMID: 25678243
18. Del Manso M , Rota MC , Declich S , Giannitelli S , Nacca G , Bolli M , et al. per il Gruppo di lavoro INFLUNET . INFLUNET: sistema di sorveglianza sentinella delle sindromi influenzali in Italia . Rapporto sulla stagione influenzale 2014 ± 2015 . Roma: Istituto Superiore di Sanità; 2016 . ( Rapporti ISTISAN 16 /18) [Internet]. [cited 2016 Nov 14 ]. http://www.iss.it/binary/publ/cont/16_18_web.pdf
19. R Core Team ( 2016 ). R: a language and environment for statistical computing . R Foundation for Statistical Computing , Vienna, Austria [Internet]. [cited 2016 Nov 14 ]. https://www.r-project.org/
20. Nunes MS with contributions from T , Heuer C , Marshall J , Sanchez J , Thornton R , Reiczigel J , et al. epiR: Tools for the Analysis of Epidemiological Data [Internet] . 2016 [cited 2016 Nov 14 ]. https://cran.rproject.org/web/packages/epiR/index.html
21. Schwarzer G. meta: General Package for Meta-Analysis [Internet] . 2016 [cited 2016 Nov 14 ]. https:// cran.r-project.org/web/packages/meta/index.html
22. DerSimonian R , Laird N. Meta-analysis in clinical trials . Control Clin Trials . 1986 ; 7 ( 3 ): 177 ± 88 . PMID: 3802833
23. Hammond GW , Cheang M. Absenteeism among hospital staff during an influenza epidemic: implications for immunoprophylaxis . Can Med Assoc J . 1984 ; 131 ( 5 ): 449 ± 52 . PMID: 6467117
24. Yassi A , Kettner J , Hammond G , Cheang M , McGill M . Effectiveness and cost-benefit of an influenza vaccination program for health care workers . Can J Infect Dis J Can Mal Infect . 1991 ; 2 ( 3 ): 101 ± 8 .
25. Nguyen-Van-Tam J , Granfield R , Pearson J , Fleming D , Keating N. Do influenza epidemics affect patterns of sickness absence among British hospital staff? Infect Control Hosp Epidemiol . 1999 ; 20 ( 10 ): 691 ±4. https://doi.org/10.1086/501568 PMID: 10530649
26. Chan ALF , Shie H-J , Lee Y-J , Lin S-J . The evaluation of free influenza vaccination in health care workers in a medical center in Taiwan . Pharm World Sci PWS . 2008 ; 30 ( 1 ): 39 ±43 https://doi.org/10.1007/ s11096-007 -9137-8 PMID: 17602309
27. Huang X , Mengersen K , Milinovich G , Hu W. Effect of weather variability on seasonal influenza among different age groups in Queensland, Australia: A Bayesian spatiotemporal analysis . J Infect Dis . 2017 ; (12) . https://doi.org/10.1093/infdis/jix181 [Epub ahead of print] PMID: 28407143
28. Ming RX , Liu JM , W Cheung WK , Wan X . Stochastic modelling of infectious diseases for heterogeneous populations . Infect Dis Poverty . 2016 ; 5 ( 1 ): 107 . https://doi.org/10.1186/s40249-016-0199-5 Erratum in: Infect Dis Poverty. 2017 Feb 28 ; 6 ( 1 ): 50 . PMID: 28003016