France has twice taken part in the WHO-PATH project. The first phase was implemented in 2004 and the second in 2007. The WHO-PATH conceptual model integrates 6 hospital performance dimensions: "clinical effectiveness", "efficiency", "human resources", "responsive governance", "safety" and "patient centeredness".
All the French hospital facilities participating in the PATH project that had data on these two indicators were included in the analysis. The sample comprised the 25 facilities that could provide complete data for these two dimensions out of the 48 short-stay participating hospitals (teaching, general and privately managed facilities) included in a voluntary basis. These 25 hospitals seemed to have the same key characteristics that the others in terms of number of beds (median was 138 [38-1670] versus 122 [15-689]--p = 0.04, type of facility (public or private) (p > 0.05) and region of origin (p > 0.05)).
Satisfaction indicators for hospitalised patients
According to OMS-PATH project, a retrospective cross-sectional postal survey was implemented on a sample of 100-250 patients per hospital facility according to size; all patients complying with inclusion criteria received a questionnaire, as they were included.
The patients included had left the facility 2 weeks to 1 month preceding the dispatch of the letter. The evaluation concerned the last hospital facility frequented by the patient. Questionnaires were sent out in February and March 2008. The patients completed the questionnaires and returned them directly to the PATH coordination unit in Nantes teaching hospital for data capture and analysis.
Hospitalisation for at least 2 consecutive nights in adult short-stay facility (medicine, surgery, obstetrics) concerning patients of 18 or over
Patients resident in France, and returning directly to their usual place of residence on discharge from the short stay unit, including homes for the elderly
Patients hospitalised for less than 2 consecutive nights in a short-stay facility, or hospitalised in rehabilitation, long-term care or psychiatry units
Patients transferred to another facility, or patients who died during hospitalisation
Patients hospitalised anonymously or confidentially, or homeless individuals
Patients declining to respond to the questionnaire or unable to do so
The three satisfaction scores for hospitalised patients belonging to the French-language scale EQS-H ("Echelle de Qualité des Soins en Hospitalisation"), validated in the literature, were calculated :
a satisfaction score relating to quality of medical information (MI)
a satisfaction score relating to relationships with staff and daily routine (RS)
a global satisfaction score
The EQS-H scale comprises 16 items divided between two dimensions: MI (8 items) and RS (8 items). The validation study on this scale showed excellent validity and stability : the first two factors explained 66% of the variance, and Cronbach's alpha coefficient for the overall scale was 0.95. A confirmation study was conducted, and similar psychometric properties were found: the two factors explained 67% of the variance and Cronbach's alpha coefficient for the overall scale was 0.93.
Five response choices are provided: "poor", "average", "good", "very good" and "excellent". Scores are attributed to each response choice (0-25-50-75-100), with higher values corresponding to greater satisfaction. Individual scores are calculated for all patients who responded to at least half the items plus one in a dimension. The scores are calculated by summing responses to items and then dividing by the number of items completed. The mean score of a dimension is the sum of individual scores divided by the number of respondents concerned. Satisfaction scores range from 0 to 100.
The patient characteristics used as adjustment variables for the scores were age with a threshold at 65 years, gender, and general satisfaction with life, coded from 1 to 7 [9, 19].
Indicators for absenteeism among nurses
Absenteeism corresponded to failure by staff to present in accordance with planned duty hours. The rate of absenteeism of short-term was defined by the WHO-PATH project as the sum of days off work on medical grounds, relating to periods of up to 7 consecutive days of sick leave, multiplied by 100, and divided by the number of equivalent full-time posts on the payroll, multiplied by 365 days. The period of the study was the first semester 2008.
One-day absences without justification were not counted, nor were days of absence for vacation and other forms of special leave, for training courses or other absences for professional reasons, or for maternity leave.
All qualified nurses in permanent posts were included.
Characteristics of the hospital facilities
The descriptive characteristics of the hospital facilities were their size, described in terms of the number of beds in medicine, surgery and obstetrics, the type of structure: public or privately managed, their French regions, their absenteeism rates and their training expenditure rates (training expenditure among all hospital facility expenditure).
Univariate descriptive statistics
To describe the characteristics of the sample, frequencies, means, standard deviation and range were calculated. Because of the small number of hospital facilities involved and the absence of normality in distributions, non-parametric statistics (Spearman's correlation test and Wilcoxon's means comparison test) were used for results according to facility (n = 25).
Multivariate statistics: multi-level model
The patient observations were grouped into clusters of hospital facilities. A multi-level model was constructed to take account of this hierarchical data structure and the hospital facility effect.
Relationship between patient satisfaction and absenteeism among nurses
Short-term absenteeism among nurses was significantly correlated with the number of hospital beds and with the type of facility. The qualitative variable "public or privately-managed facility" was included in the model, as well as three patient characteristics: age with a threshold at 65 yrs, gender, and satisfaction score for life in general (median threshold: 4).
Individual level equation for patient i in hospital facility j
Hospital facility level equations
Scoreij is the value of the score patient i in facility j among the 25 hospital facilities having taken part in the satisfaction survey and having collected data on absenteeism among nurses.
βagei is regression coefficient of the age fixed effect at the individual level.
xageijis the fixed effect variable for patient i in hospital facility j at the individual level.
is the error for patient i in hospital facility j at the individual level.
βabsenteeism is the regression coefficient of the absenteeism fixed effect at the hospital facility level, it is identical for all groups.
xabsenteeismjis the fixed effect variable in hospital facility j at the hospital facility level.
u0jis the error in hospital facility j at the hospital facility level.
The significance threshold is set at 5%.
The data analysis was performed on S-PLUS 6.0 and R 2.9.0 software.