Dr. Ghazi M Al Qatari, M. Comm. H., PhD. (1)
Dr. Dave Haran, Msc., PhD. (2)

1. Department of Primary Health Care, Qateef, Eastern Province, Saudi Arabia.
2. Department of Health System Development, Liverpool School of Tropical Medicine, Liverpool, UK.

Correspondence to:
Dr. Ghazi M Al Qatari,
P.O.Box 709, Qateef 31911, Saudi Arabia.
Tel: +966 3 8551525. Fax: +966 3 8541589.
E-mail: ghazialqatari@gmail.net
.........................................................................................................................

Keywords: patient satisfaction, primary health care, quality assurance, structure, utilisation.
..........................................................................................................................

Evaluation of the quality of health care has emerged as a key issue for all health services, and for some time it has been recognised that the patient's views are essential components of such evaluations. Social acceptability and patient satisfaction are now acknowledged as important criteria for the evaluation of health services1. Patient satisfaction is recognised as a valuable outcome in itself and, like any outcome, requires measurement2. Investigations of patients' satisfaction have used a range of interview and questionnaire measures which differ quite markedly in their levels of methodological sophistication3. The simplest method of all has been to use general questions about the topics of interest. Items with fixed multiple choice response categories have a number of advantages for statistical purposes4,5. The use in questionnaires of a Likert-type format for such items, where responders are asked to indicate the extent of their satisfaction with certain aspect of services is another method that fulfils the same objective6. Measurement of satisfaction, although important and highly recommended, should not be the ultimate objective. The point is not simply to measure quality but to improve it7. This is to say that finding out about the most likely determinants of satisfaction is an important step if a real improvement in the health care service or setting under question is to be made. Many studies sought to measure the levels of satisfaction only, while some others went further to reveal the factors determining it but failed to specify the most influential factor on the levels of satisfaction measured. The recognition and the analysis of such factors are fundamental in modern thinking about quality improvement8,9,10,11. This study, like the population satisfaction study11, is a small step towards this direction. Studies that take this approach using a wide range of variables will contribute to real improvement in the health care services and settings in Saudi Arabia.

Patients
This study was conducted concurrently with the population satisfaction study11 on April 1995. It is thought that this will give a more complete picture about satisfaction than conducting either alone. Many studies showed that the setting in which the people are interviewed plays an important role in their responses especially about their satisfaction with health care 3,12,13. All patients attending the 25 primary health care centres in Qateef area during the week of the data collection for each centre were viewed as the target population. In practical terms, any file that was pulled from the medical records' room for patients who were 15 years or older would be part of the sampling frame for the target population. For the patients less than 15 years of age interviews were made with their older companions and it was the companions' opinions that was sought. If there was more than one companion the older was interviewed and for couples with their children the mother was interviewed as she was the one expected to know more about the child and most likely to accompany the child to receive the required services in the centres. Patients whose illness prevented them from participating in the study were excluded.

Sampling
The sample size was estimated using the same strategy as used in the population study11 so that valid comparisons can be made. A 95% confidence interval of +/- 5% for proportions was selected. The required sample is 400. To allow for age or sex breakdowns of the data this number was doubled to 800. Subjects were selected using a proportional allocation method whereby systematic sampling of every nth attender at each health centre was entered into the study. The value for n was determined from the average daily attendance at that centre. The selection process was constrained by the time taken to complete each interview. It was estimated that each interviewer would be able to interview every fifteenth eligible patient attending the health centre which means that n=15. With sufficient well-trained interviewers the data collection for the patient satisfaction survey could be completed within the same four weeks period employed in the population survey mentioned above. Within this four week period interviewing would be carried out for a period of one week in each health centre. With this strategy the number of patients interviewed at each centre would be proportional to the number of patients attending the centre during the target week. Hence the overall sample would be weighted approximately by the average utilisation of the health centre. When the selected patient declined to participate, substitution of the previous eligible patient was allowed. While it is noted that this might have introduced a bias as compared with complete random sampling it has been nevertheless done to increase the efficiency of sampling. The final sample size reached was 1098.

Instrument
The process of questionnaire construction, testing and revision for the patient survey was similar to that for the population survey11. The questionnaires were also subjected to similar pre-testing and piloting experiences with exactly similar results. One major difference from the questionnaire in the population survey is that the questionnaire form has two main parts - an entry part that is completed as the client enters the health centre and the other is an exit part that is completed as the patient was about to exit the health centre. Another difference is the presence of "nature of health centre visit" as a variable that is present in this study but not the population study. Like in the population study the components and their items had been chosen by the authors based on experience and literature review.

Because of traditional / religious reasons same-sex interviewers were required for the male and female respondents. Whilst this increased the number of interviewers required, and hence the possible variability of the way interviews were conducted, it also had the advantage of reducing embarrassment and possible inhibitions that females might have in answering questions to male interviewers (and vice versa).

Female patients were interviewed by well-trained health visitors (HVs) and male patients interviewed by well-trained health inspectors (HIs). The 25 PHCCs were covered in four weeks. The selection of HVs and HIs was based both on their willingness to undertake the training and interview work, and on judgements as to their suitability for interview work. Training was similar to that carried out for the interviewers in the population survey. As the proportion of non-Saudi patients was low and as the great majority of them were Arabic speakers no language problems had been encountered.
Data processing of questionnaires as well as the analysis plan for the patient survey followed the same strategy as that for the population survey.

Cronbach's alpha test in the (SPSS) was conducted to test the level of internal consistency and so the reliability of the components with multiple items. Like other correlation coefficient scores close to 10.00 are highly correlated and scores close to 0.00 are not correlated. Typically a minimum standard for group comparisons for reliability of a measure is above 0.5 on the Cronbach's Alpha measure. Scores above this suggest high reliability. Nevertheless, there is one point that should be taken into consideration. The Alpha measure will increase as the number of items used in every component increases. Therefore for areas using fewer items Alpha score may be lower.

Table 1 shows the coefficient alpha scores to be generally high to moderate. Of all the components attitude and to a lesser extent explanation received low Cronbach's value. All this means that we should be cautious in interpreting the ability of the questionnaires used to measure satisfaction with these components in a consistent and reproducible fashion.

Like in the population study, the questionnaires were subjected to face, consensual and construct validity.

Table 2 shows the frequency distribution of the various independent variables among the respondents in the study.

Table 3 shows the satisfaction rate received by each of the components. Waiting area structure, environmental structure, and explanation received the lowest levels of satisfaction and hence are priorities for improvements. Patients were most satisfied with the staff attitude, perceived outcome of the consultation episode, and consultation time.

Table 4 shows the extent of significance of the relationship between each of the various socio-demographic / independent factors and the components used in the study as well as the Beta values reflecting the relative importance of the associations between them. It is evident from the table that some of the components have a significant relationship with more than one factor. It is to be realised that the identified areas of concern are mostly associated with the extent of PHCC use, type of PHCC building, age and literacy status.

It is clear from table 4 that the relation between satisfaction levels and the various independent factors were not following a simple one to one cause and effect mechanism and the effect is more likely to have been brought about by a number of causes, some inter-related while some others are independent of each other.

In this study, each of the components were entered in a multiple regression analysis test against those independent variables and the resulting beta values were used to identify the relative importance of the various independent variables used. Nevertheless, the low coefficient scores in the Cronbach's alpha test for attitude and to a lesser extent explanation does not allow us to generalise our conclusions to all components and limit it only to the other four components. As far as the two structure components are concerned, the test showed that the type of the PHCC building was the most influential factor followed by the literacy level and the sex of the respondent respectively. As for activities age was the only influential factor whereas the extent of health centre use or how regular the respondent was the only influencing factor as far as consultation time is concerned. In general, the extent of health centre use did not show high beta values although its effect on the various components was consistent.

The type of the PHCC building was the factor most associated with patients' satisfaction with both structure of the waiting area and the environment structure. Of those using purpose-built PHCCs 42% were satisfied with the waiting area structure as compared with a satisfaction rate of 68.4% among those using rented PHCCs. Similarly, 61.6% of those using purpose-built PHCCs were satisfied with the environment structure as compared with a satisfaction rate of 79% among those using rented PHCCs. The factor with the third highest beta value was literacy level against the waiting area structure. The satisfaction rate among the literate was 56.1% as compared to 73.3% among the illiterates. Sex was the factor that showed the fourth highest beta value against the waiting area structure with 59.1% satisfaction rate among females as compared to 65.3% among males.

The results in many satisfaction studies revealed that satisfaction is multi-factorial, and no one factor could be claimed to be the sole contributor to satisfaction or dissatisfaction14. Nevertheless, some factors are more important than others in contributing to patient satisfaction. Identifying the relative importance of the variables helps to rationalise decisions related to the improvement of health care so that they are not limited to satisfaction rates only. This is a clear message to the young primary health care quality assurance programme in Saudi Arabia as it is hoped that it will ultimately incorporate users' satisfaction elements in such programmes. The same could also be said about any quality assurance programme in places other than primary health care.

This study as well as some other studies used multiple regression analysis to reveal the relative importance of the independent variables used. The fact that the coefficient alpha was low with attitude and explanation components makes it difficult to generalise our conclusions regarding what independent factor is more influential than the other and leads us to limit it to the other components that received moderate to high coefficient alpha.

The study showed that the type of the primary health care centre building had the strongest influence on the level of satisfaction with higher beta values against the environment structure and more so against the waiting area structure. This simply tells us that the setting in general has an important effect on satisfaction and care delivered in a reasonably accepted setting is most likely to be accepted. To our knowledge, no other Saudi study tried to see the effect of the nature of the building on satisfaction. The study also showed that, with the reservations mentioned above as regards the low coefficient alpha values for attitude and explanation components, the extent of health centre utilisation did not appear to be as important in this study as in the population satisfaction study11. The literacy status was the second strongest factor influencing satisfaction with structure components with the illiterate patients being more satisfied than the literate. Other Saudi studies found variable effects of literacy state on the level of satisfaction. Whereas Al Faris15 did not find any influence of the educational level, Makhdoom et al16 found, like in this study, low educational level significantly associated with satisfaction. Other studies found variable results with possible effects from other socio-demographic factors14,17,18. Sex of patient was also an important influence on the level of satisfaction. Males were more satisfied with both structure components than females. Other Saudi studies showed variable sex effects. Saeed et al19 and Al Faris15 did not find any significant association between the level of satisfaction and sex, while Al Dawood20 identified maleness of the respondent as the most influential variable on the level of satisfaction. Other studies also showed variable and inconsistent effect of sex on the level of satisfaction21. While some showed that women give more negative responses than men22,23, some others found females to be more satisfied than males17,24,25. Nature of health centre visit had a relatively weak effect on the component of activities. No Saudi study included factors like the type of the building, the nature of the area in which the respondent lives, the extent of the PHCC use or nature of the visit in their independent variables. Such factors might affect the expectations the respondents have for their centres and so could influence their level of satisfaction with its settings and services. The finding in this study that the type of the PHCC building was the most influential factor on the level of satisfaction should be put in mind in any future quality improvement effort. Users of PHCCs with purpose-built buildings were found to be less satisfied than the users of centres with rented buildings. This matches findings from the population study11. Although this finding could be explained by the high expectation that people might have for the PHCCs with purpose-built buildings, the topic needs to be further studied in both settings and in various locations.

We thank all the health visitors and the health inspectors who actively participated in interviewing patients and whose perseverance enabled us to complete the health centre study smoothly. The help and understanding of the various staff in the health centres is quite appreciated. We also thank all the patients who participated in the study for their co-operation and understanding.