International Journal for Quality in Health Care Advance Access originally published online on September 14, 2007
International Journal for Quality in Health Care 2007 19(6):382-389; doi:10.1093/intqhc/mzm041
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Evidence of a non-linear influence of patient age on satisfaction with hospital care
1 Public Health Department, University Hospital of Nantes - France
2 Inserm, U669, Paris, F-75014 France ; Univ Paris-sud 11, Le Kremlin Bicêtre, F-94000 France ; Univ Paris 5, Paris, F-75015 France ; APHP, Villejuif, F-94804 France
3 Direction de la Politique Médicale, APHP, Paris, France
Address reprint requests to: Leila Moret; Dr. Leïla Moret Unité Qualité-Risques-Evaluation Public Health Department - PIMESP, Hôpital Saint-Jacques, 44093 Nantes CEDEX, Tel: +33 2 40 84 69 24; Fax: +33 2 40 84 69 21; E-mail: lmoret{at}chu-nantes.fr
 |
Abstract |
|---|
 Top Abstract Patients and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
Background. Patient age is the main socio-demographic factor influencing patient satisfaction with care but the nature of the relationship between age and patient satisfaction is controversial.
Objective. This study aims to clarify whether the association of age with satisfaction is linear or shows some other configuration.
Methods. Data were obtained from two different satisfaction studies conducted in 27 short-stay teaching hospitals. Study 1 included 1547 inpatients, who completed the EQS-H questionnaire at the time of discharge. Study 2 included 7624 inpatients interviewed by phone at home after discharge, who answered the SAPHORA questionnaire. On the basis of the results of the exploratory analysis, three models for adjustment of age on satisfaction were compared: a simple linear model, a five-group step function and a linear model with a change in slope.
Results. The most suitable model for adjusting patient age to satisfaction scores for quality of medical and nursing care, whether for the EQS-H or the SAPHORA scale, was not a linear relationship: patient age was linearly and positively correlated to satisfaction before 65 years and negatively thereafter. Adjustment of patient age to accommodation and premises satisfaction scores proved to be different, closer to a linear relationship.
Conclusion. These results suggest considering the patient age variable as a non-linear factor for adjusting satisfaction scores, in particular in relation to care. Further studies are needed to confirm the evidence of a threshold around 65 years beyond which satisfaction scores for the quality of medical and nursing care decrease.
Keywords: adjustment, inpatient satisfaction, patient age, questionnaire, quality of care
The assessment of satisfaction with care is considered as a major component of quality management and reflects healthcare professionals' ability to meet their patients' needs and expectations. Numerous studies on patient needs and expectations have been conducted and many patient satisfaction scales have been developed and validated. The effect of clinical characteristics and socio-demographic factors on satisfaction scores has been widely studied, but their impact on satisfaction is still unclear. Indeed, results are inconsistent and diverge from one study to another [1, 2]. According to the literature, gender, ethnic origin, educational, social, or marital status and conditions of admission, have all been found to be associated [3–6]. Subjective health status also seems to affect the satisfaction score: subjects who believe they were in poor health before hospitalization tend to be less satisfied, as do those who consider that their health has not improved by the end of hospitalization [4, 6–9]. However, the effect of subjective health status seems to be smaller after adjustment for all other co-variables [10]. In a meta-analysis carried out in 1988, Hall [11] concluded that ‘socio-demographic characteristics are a minor predictor of satisfaction, at best’.
The only factor almost always associated with satisfaction score is patient age [1, 2, 11]. Most authors agree on the general nature of the relationship: the younger patients are less satisfied than the older patients [2, 8, 12–15]. All of these studies show either a positive linear relationship between the satisfaction score and patient age [2, 16, 17], or different average satisfaction scores for each age group [3, 4, 18, 19]. Conversely, Kane showed a constant negative relationship between age and satisfaction [20] and in 2000, Boudreaux did not find any significant relationship between age and overall satisfaction [21]. Recent studies, however, suggest a more complex relationship between patient age and satisfaction scores [18, 22]. A drop in satisfaction scores beyond a certain age threshold—which varies, depending on the author, between the ages of 65 and 80—has been observed by Jaipaul and Lee [23, 24]. In these two studies, age is considered to be a piecewise constant function, and mean satisfaction scores for each age group are compared.
Although these studies suggest the need to adjust satisfaction scores according to age, the present work aims to clarify the profile of the age–satisfaction relationship, and to determine whether the association of age with satisfaction is linear or possesses some other configuration. To answer this question, we compared several models on the basis of the explanatory analysis. This method was applied to two different hospitalized patient samples on data obtained from two different validated satisfaction scales widely used in hospitals.
|
Patients and methods |
|---|
|
TopAbstractPatients and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
Study setting and participants
Study 1
Data were collected in two stages in 2002 and 2003 from 1547 inpatients hospitalized in a big teaching hospital. All patients aged 18 and over, able to complete a self-rated questionnaire, hospitalized full-time for at least 24 h in 14 medicine, surgical and obstetrics wards, were included consecutively. Children, patients not able to answer (for reasons of health status or language barrier) and outpatients were excluded. Patients were approached the day before discharge by independent, trained and research assistants. They explained the purpose of the study, invited patients to take part and gave them a questionnaire. Patients completed the anonymous questionnaire alone, and returned it to the assistants in a sealed envelope. A total of 106 patients refused to participate for the two data collection stages (5.6%). Further to this, 151 patients (8.4%) were not available at certain moments (consultation, investigations, care procedures and family visits).
The widely used EQS-H scale assesses inpatient satisfaction with medical and nursing care including items related to medical information provided and relationships with healthcare providers. Each of the 16 items is rated on a five-point scale ranging from 0 to 4 (poor, fair, good, very good and excellent). The instrument is uni-dimensional and has shown excellent psychometric properties (Cronbach 
0.95).
Study 2
Data, derived from 26 short-stay teaching hospitals in the Paris region, were collected in 2001–3 from 7624 inpatients. All patients aged 18 and more, hospitalized in medicine, surgery, obstetrics and psychiatry units were contacted over the phone 2 weeks after discharge. Children, patients not able to answer (on account of health status or language barrier) and outpatients were excluded.
For this purpose, a list of the last 250 patients discharged and meeting inclusion criteria was drawn up from administrative files. Patients were then randomly selected from the list, which was exploited in a progressive manner. Interviews were conducted by professional interviewers trained by a polling company. To optimize participation rates, patients were called up to 12 times on the phone on different days (including Saturdays) and at different times (including evenings). Each patient that could not be contacted was serially replaced by another.
The overall average non-participation rate for the three data collection stages was 5% (1.5% of patients refused to participate, and the other two main reasons for non-participation were death after the hospitalization (0.7%) and re-hospitalization (2.8%)), these being a posteriori exclusion criteria. Twenty-nine percent of the patients could not be contacted by phone (no phone number available, wrong number and no reply after 12 attempts).
Patients in this second study answered another widely used validated satisfaction questionnaire (SAPHORA) [25], exploring medical and nursing care, organization of discharge, and accommodation and premises. The SAPHORA scale comprises 26 items in 3D: 3 dimensions medical and nursing care (14 items), medical and administrative organization of discharge (4 items) and accommodation and premises (8 items). Each item is rated on a five-point scale ranging from 0 to 4 (poor, fair, good, very good and excellent). The instrument has good psychometric properties (all Cronbach coefficients are over 0.80).
Data analysis
Calculation of the overall satisfaction scores
According to the specifications of the instruments [19, 25], scores were calculated for each patient as follows: sum of the values obtained for each item in a dimension, divided by the number of completed items. A score was calculated when at least half the items were completed. Scores were then expressed on a scale from 0 to 100.
Analysis of the relationship between patient age and satisfaction scores
Explanatory analysis
Box-plots. Five age groups were formed according to the widely used classifications (18–34, 35–49, 50–64, 65–79 and 80 years and over). Box-plots for the satisfaction scores derived from the two scales in relation to patient age groups in years were provided so as to visualize the form of the data.
Comparison of mean scores according to age group. An ANOVA procedure enabled comparison of mean satisfaction scores for each age group.
Determination of the best model for the relationship between patient age and satisfaction
On the basis of the descriptive results obtained, three models were compared for the purpose of relating age to satisfaction data.
Simple linear regression. A linear regression was computed taking the variable ‘age’ to be a continuous linear variable, so as to evidence any increasing or decreasing linear relationship with the satisfaction score.
Linear regression with age as a continuous variable in two slopes. A linear regression model with age as a continuous variable in two slopes with the threshold at 65 years was computed.
Step function with five age groups. A linear regression was computed with the same five age groups as in the ANOVA (18–34, 35–49, 50–64, 65–79 and 80 years and over). The reference class was 18–34 age group.
Selection criterion for the best adjustment model for patient age data on satisfaction. To compare the three models, one with the other, for each satisfaction score the Bayesian Information Criterion (BIC) was calculated [26]. This criterion is one of those that are widely used to compare simultaneously the statistical quality of a model (the fit) and its conceptual quality. It is thus also known as the parsimony index. In practical terms, the comparisons are not based on statistical tests, but on the choice of the model that optimizes the BIC. The model selected is that for which the BIC showed the lowest value. To test the robustness of the results, a cross-validation was conducted: three random selections of 50% of the sample three times in succession were done.
In all analyses, p < 0.05 was considered statistically significant. All analyses were performed using S-PLUS software application (sixth version).
|
Results |
|---|
|
TopAbstractPatients and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
Profiles of respondents and non-respondents (Table 1)
The patients who refused to participate were significantly older than the mean in both samples (p < 0.001). In addition, the patients who could not be re-contacted were significantly older than overall participants in sample 1 (p < 0.01), whereas this difference was not significant in sample 2.
|
Characteristics of patients and satisfaction scores (Tables 1 and 2)
Mean age of the respondents was five points lower in sample 2 than in sample 1 (Table 1).
|
The quartiles were 34, 56 and 75 years in sample 1 and 34–49–62 years in sample 2. A half and a third of patients were admitted in emergency in the two samples, respectively. In sample 2, more than the half underwent surgery during hospitalization, and only 43.4% in sample 1. The average length of stay was higher in sample 2 (Table 2).
Analysis of the relationship between patient age and satisfaction scores
Explanatory analysis
Box-plots (Fig. 1). For the scores that relate to care (Fig. 1a and b), the graphic analysis appeared to show up the existence of an alteration in the relationship beyond 65 years of age: the score increases before this threshold, and decreases thereafter. Fig. 1c (score for ‘organisation of discharge’ which comprises only four items) was not really informative in graphic form, showing a score decreasing after 80 years. Finally Fig. 1d (score for ‘accommodation and premises’) seemed to be more generally in favour of the existence of a continuously increasing relationship between patient age and satisfaction.
|
Comparison of mean scores according to age groups (Table 3). For scores relating to the quality of medical and nursing care, an increase in mean scores was observed as far as the 50–64 age group, and a decrease thereafter. Table 3 suggested the existence of a threshold beyond which mean satisfaction scores begin to decrease. Mean scores per age group below and above the threshold were significantly different for both instruments. Similar results were observed for the ‘medical and administrative organization of discharge’ scores obtained from SAPHORA. For the ‘accommodation and premises’ dimension, the mean SAPHORA score peaked in the 65–79 age group.
|
Comparison of three models for adjustment of age on satisfaction (Table 4)
|
Simple linear regression. A non-significant relationship was found between patient age and EQS-H scores. Conversely, a significant relationship was observed between patient age and SAPHORA scores but the correlation was not strong (very low coefficients), sometimes positive and sometimes negative.
Linear regression with age as a continuous variable in two slopes. The results of the linear regression with one threshold in the slope at 65 years yielded similar results for the two scores relating to care (EQS-H and SAPHORA ‘medical and nursing care’). The regression coefficients were significantly positive before 65 years and significantly negative afterwards. For the score ‘organisation after discharge’, age was significantly negatively linked to the score after 65, while the relationship between age and satisfaction in relation to ‘accommodation and premises’ appeared to increase in a linear manner before 65, and not to be significantly linked thereafter.
Linear regression with five age groups. For satisfaction scores relating to care (EQS-H and SAPHORA ‘medical and nursing care’), in relation to the reference age group (18–34), the mean satisfaction score was significantly higher (>4 points) for the 50–64 age group. The score remained significantly higher, but falling off, for the 65–79 age group. This decrease then became more marked, the difference becoming non-significant or barely significant in relation to the reference age group. For the score ‘organisation after discharge’, a fairly similar phenomenon was observed, with lower coefficients. Finally, for the ‘accommodation and premises’ score, the coefficients were still positive and significantly higher than in the reference age group.
Choice of the best model, i.e. the model for which BIC value was minimum. The best adjustment of age data on satisfaction in relation to care (EQS-H and SAPHORA ‘medical and nursing care’) was obtained using the 2-slope linear regression model, with the threshold occurring at the age of 65. This model was also the best for adjusting age on the satisfaction score relating to ‘organisation of discharge’.
Conversely, the data relating to accommodation and premises were better adjusted on a single-slope classic linear regression.
Cross-validation confirmed all the results obtained, showing that adjusting age data to satisfaction with care and to satisfaction with organization of discharge was better achieved when age was considered as a continuous variable with a threshold in the slope at 65 years. Results were also stable for data relating to accommodation and premises.
|
Discussion |
|---|
|
TopAbstractPatients and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
Our study shows that the best fit of patient age to satisfaction data is not a linear relationship, in disagreement with results described in the literature [2, 16, 20, 21]. The present results confirm and refine those reported by Jaipaul [23], using satisfaction with care scores in 5D: dimension (physician care, coordination of care, nursing care, discharge instructions and information provided). The author also demonstrates the existence of a threshold at 65 years, beyond which the satisfaction score decreases. Further to this, several authors have highlighted a decrease in satisfaction scores relating to care after 65 years [18, 22]. In the present study, similar results have been shown with the score for the ‘organisation of discharge’. However, these results are not so clear-cut, since they are derived from a set of only four items. Finally, if the satisfaction score for ‘accommodation and premises’ is considered, results are different from those obtained for the medical and nursing care dimension. Here the relationship appears closer to an increasing linear relationship, suggesting that younger patients have higher expectations than the others. Several limitations have to be considered. First, these results only concern hospitalized patients and cannot be generalized to outpatients. Second, the older respondents could experience more difficulty than younger patients in completing questionnaires, and the non-response bias could have an effect. Gasquet en 2001 [27], has shown that reminders made it possible to improve response rates for satisfaction questionnaires, but that the satisfaction scores obtained in this way were not significantly different according to whether or not the patient responded spontaneously or after a recall. In the present study, the rates of non-response are similar to those found by other authors [28, 29]. A further limitation is that the only comparisons possible were of for medical and nursing care dimensions, using two different scales. Two other domains of patient satisfaction (accommodation and premises and organization of discharge) were each explored by only one of the scales and comparison was therefore not possible. Moreover, other domains of patient satisfaction could also have been explored.
The results obtained are interesting in that identical patient age adjustment models were obtained on two different, validated and widely-used medical and nursing satisfaction scales, using different data collection methods, on two large inpatient samples from several different hospitals in two distinct regions (one mainly urban and the other mainly rural). Identifying the best adjustment model of patient age on satisfaction should make it possible to reduce bias in analyses, in particular in comparing scores between different departments or hospitals. The age structure of patient populations can differ between wards and hospitals, depending upon their care offer. Thus, taking patient age into account in analysing satisfaction data is a necessary part of clinical benchmarking, and the follow-up of hospital quality procedures.
Why are satisfaction scores obtained from the younger patients significantly lower than those obtained from patients of 50 or 60? Very few arguments are put forward in the literature. Breemhaar [16] recalls that the ‘elderly and younger patients are treated differently by hospital staff, and professional workers find it more difficult to supply information to elderly cancer patients’. Jaipaul has suggested that, ‘older patients may be conditioned to a paternalistic model of care and may find current models, which expect more active patient participation, to be overwhelming and intimidating. Younger patients are likely to expect to be involved in the decision-making process and would most probably find any alternative unacceptable’ [23]. In addition, the expectations of patients may not be the same at different times in life, and growing introspection among elderly patients certainly makes communication more difficult. Finally, there is the issue of the relevance and the interpretation of satisfaction questionnaires used to survey elderly patients. And a question remains: are the tools used to collect data on satisfaction with care suited to elderly patients?
In conclusion, our results suggest that age is an important modifier of satisfaction and should be taken into account when interpreting these kinds of data. The patient age variable should be considered as a non-linear factor for adjusting patient satisfaction scores.
|
Funding |
|---|
|
TopAbstractPatients and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
|
Acknowledgement |
|---|
|
TopAbstractPatients and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
The authors would like to thank Angela Swaine Ven dier for help with the drafting of the english.
|
References |
|---|
|
TopAbstractPatients and methodsResultsDiscussionFundingAcknowledgementReferences |
|---|
- Sitzia J, Wood N. Patient satisfaction: a review of issues and concepts. Soc Sci Med (1997) 45:1829–43.[CrossRef][Web of Science][Medline]
- Hall JA, Dornan MC. Patient sociodemographic characteristics as predictors of satisfaction with medical care: a meta-analysis. Soc Sci Med (1990) 30:811–8.[CrossRef][Web of Science][Medline]
- Nguyen Thi PL, Briancon S, Empereur F, et al. Factors determining inpatient satisfaction with care. Soc Sci Med (2002) 54:493–504.[CrossRef][Web of Science][Medline]
- Jackson JL, Chamberlin J, Kroenke K. Predictors of patient satisfaction. Soc Sci Med (2001) 52:609–20.[CrossRef][Web of Science][Medline]
- Wallin E, Lundgren P, Ulander K, et al. Does age, gender or educational background affect patient satisfaction with short stay surgery? Ambul Surg (2000) 8:79–88.[CrossRef]
- Cohen G, Forbes J, Garraway M. Can different patient satisfaction survey methods yield consistent results? Comparison of three surveys. BMJ (1996) 313:841–4.
[Abstract/Free Full Text] - Cleary PD, McNeil BJ. Patient satisfaction as an indicator of quality care. Inquiry (1988) 25:25–36.[Web of Science][Medline]
- Williams SJ, Calnan M. Convergence and divergence: assessing criteria of consumer satisfaction across general practice, dental and hospital care settings. Soc Sci Med (1991) 33:707–16.[CrossRef][Web of Science][Medline]
- Hall JA, Feldstein M, Fretwell MD, et al. Older patients' health status and satisfaction with medical care in an HMO population. Med Care (1990) 28:261–70.[CrossRef][Web of Science][Medline]
- Linder-Pelz SU. Toward a theory of patient satisfaction. Soc Sci Med (1982) 16:577–82.[CrossRef][Web of Science][Medline]
- Hall JA, Dornan MC. Meta-analysis of satisfaction with medical care: description of research domain and analysis of overall satisfaction levels. Soc Sci Med (1988) 27:637–44.[CrossRef][Web of Science][Medline]
- Larsen DE, Rootman I. Physician role performance and patient satisfaction. Soc Sci Med (1976) 10:29–32.[Medline]
- Henley B, Davis MS. Satisfaction and dissatisfaction: a study of the chronically-ill aged patient. J Health Soc Behav (1967) 8:65–75.[CrossRef][Web of Science][Medline]
- Cohen G. Age and health status in a patient satisfaction survey. Soc Sci Med (1996) 42:1085–93.[CrossRef][Web of Science][Medline]
- Williams B. Patient satisfaction: a valid concept? Soc Sci Med (1994) 38:509–16.[CrossRef][Web of Science][Medline]
- Breemhaar B, Visser AP, Kleijnen JG. Perceptions and behaviour among elderly hospital patients: description and explanation of age differences in satisfaction, knowledge, emotions and behaviour. Soc Sci Med (1990) 31:1377–85.[CrossRef][Web of Science][Medline]
- Labarere J, Francois P, Bertrand D, et al. Outpatient satisfaction: validation of a French-language questionnaire: data quality and identification of associated factors. Clin Perform Qual Health Care (1999) 7:63–9.[Medline]
- Gonzalez N, Quintana JM, Bilbao A, et al. Development and validation of an in-patient satisfaction questionnaire. Int J Qual Health Care (2005) 17:465–72.
[Abstract/Free Full Text] - Salomon L, Gasquet I, Mesbah M, et al. Construction of a scale measuring inpatients' opinion on quality of care. Int J Qual Health Care (1999) 11:507–16.
[Abstract/Free Full Text] - Kane RL, Maciejewski M, Finch M. The relationship of patient satisfaction with care and clinical outcomes. Med Care (1997) 35:714–30.[CrossRef][Web of Science][Medline]
- Boudreaux ED, Ary RD, Mandry CV, et al. Determinants of patient satisfaction in a large, municipal ED: the role of demographic variables, visit characteristics, and patient perceptions. Am J Emerg Med (2000) 18:394–400.[CrossRef][Web of Science][Medline]
- Labarere J, Fourny M, Jean-Phillippe V, et al. Refinement and validation of a French in-patient experience questionnaire. Int J Health Care Qual Assur Inc Leadersh Health Serv (2004) 17:17–25.[Medline]
- Jaipaul CK, Rosenthal GE. Are older patients more satisfied with hospital care than younger patients? J Gen Intern Med (2003) 18:23–30.[CrossRef][Web of Science][Medline]
- Lee Y, Kasper JD. Assessment of medical care by elderly people: general satisfaction and physician quality. Health Serv Res (1998) 32:741–58.[Web of Science][Medline]
- Pourin C, Tricaud-Vialle S, Barberger-Gateau P. Validation d'un questionnaire de satisfaction des patients hospitalisés. Journal. d'Economie Médicale (2003) 21:167–181. (in French).
- Schwarz G. Estimating the dimension of a model. Ann. Stat. (1978) 6:461–4.[CrossRef]
- Gasquet I, Falissard B, Ravaud P. Impact of reminders and method of questionnaire distribution on patient response to mail-back satisfaction survey. J Clin Epidemiol (2001) 54:1174–80.[CrossRef][Web of Science][Medline]
- Labarere J, Francois P, Auquier P, et al. Development of a French inpatient satisfaction questionnaire. Int J Qual Health Care (2001) 13:99–108.
[Abstract/Free Full Text] - Sitzia J, Wood N. Response rate in patient satisfaction research: an analysis of 210 published studies. Int J Qual Health Care (1998) 10:311–7.
[Abstract/Free Full Text]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||