Cost-consciousness among Swiss doctors: a cross-sectional survey

Switzerland has the second highest health care expenditures per capita in the world, behind the United States, and devotes about 11% of the GNP to health care [25, 26]. Doctors in private practice who are paid on a fee-for-service schedule provide most ambulatory care, and most hospitals are public, subsidised by local governments, where doctors are salaried (in addition, some senior hospital doctors have private practice privileges). To achieve social solidarity, the Health Insurance Law makes compulsory the purchase by households of a fairly comprehensive package of health benefits, which includes ambulatory treatment, inpatient care, home nursing care, and some health promotion activities. The basic health insurance coverage can be contracted from approximately 90 private insurance carriers, which are not allowed to earn profits from the mandated benefit package. Insurers can offer different schemes for health care provision such as patients' free choice of physician ("any-willing-provider" or compulsory contracting) or preferred providers' contracts (general practitioner-gatekeeper model or restricted network of providers). Restriction on the choice of the provider results in lower premiums for the patient. According to official statistics, in 2001, about 45% of the insured chose the lowest permissible deductible and only 9% the highest; about 8% chose a managed care policy, mainly the general practitioner-gatekeeper model. Regarding the organisation of care, the Swiss health care system has therefore been qualified as a "regulated competition system without managed care" [27]. The canton Geneva has the highest health care costs and medical density in the country. In 1998, there were 58 doctors per 10000 residents in canton Geneva – 31 in private practice, and 27 in public hospitals.

We conducted a cross-sectional mail survey of doctors practicing in canton Geneva, Switzerland, during the fall of 1998. Doctors were identified from membership files of the Geneva Medical Association (1370 members) and the Swiss Association of Interns/Registrars, Geneva Section (906 members). After exclusion of 54 duplicate records, 10 pre-test participants, 97 doctors who had incorrect addresses and 121 who did not practice clinical medicine, 1994 doctors were eligible.

Attitude toward costs (cost-consciousness) was measured with a validated 6-item instrument [19]. Two items explore the doctors' opinion about health care costs in general, three items explore their attitudes regarding the costs of tests and procedures, and a last item probes the importance of out-of-pocket payments (Table 1). The items were scored on a five-point Likert scale anchored by 1: totally disagree and 5: totally agree. Negatively worded items were reversed so that a higher score would mean greater cost-consciousness. We translated the instrument into French using a standardized procedure (three independent translations, selection of a consensus translation by an expert panel, pre-tests among 10 physicians) (Appendix 1). The internal consistency of the translated scale was similar to that of the original scale (Cronbach α : 0.68, versus 0.74 for original) and a factor analysis indicated that the translated items represented a single dimension. We computed a summary cost-consciousness score whenever at least 3 of the 6 items were present.

Determinants of cost-consciousness included socio-demographic characteristics (age, sex), time since graduation from medical school, type of practice (public versus private), medical specialty, workload characteristics (number of patients per week, time spent with a new patient), stress from uncertainty, and work-related satisfaction.

Stress from uncertainty refers to physicians' reaction to their limitations of professional knowledge, problems of diagnosis, ambiguities of treatment and outcome, unpredictability of patients response, and variations in physicians' attitudes, values and perceptions of risk [28‐33]. Items of a stress from uncertainty scale [28] were translated into French using the same procedure as for the cost-consciousness scale. Factor analysis confirmed that the items represented a single dimension. The internal scale consistency coefficient was 0.88.

Work-related satisfaction was measured by a 17-item instrument based on the work of the Society of General Internal Medicine Career Satisfaction Study Group [34], containing questions addressing satisfaction with workload, intellectual stimulation, leisure time, work stress, relations with patients, autonomy in treating patients, autonomy in specialist referrals, overall quality of care, relations with peers, relations with nurses and other non medical staff, administrative burden, continuing medical education, enjoyment of work, respect and prestige, payment mechanism, current income, and job satisfaction overall [35]. The internal scale consistency coefficient of a work related satisfaction scale including all 17 items was 0.88.

Descriptive statistics were calculated for each item of the cost-consciousness instrument (mean, standard deviation, floor & ceiling effect). Summary scores were calculated by averaging the responses to the relevant items, whenever half or more were present. For comparisons with the original [19], the cost-consciousness score was rescaled between 0 and 24; to facilitate the interpretation of regression models, we also standardized this score to mean 50 and standard deviation 10.

We explored the relationships of the cost-consciousness scale with socio-demographic and job characteristics of the respondents, stress from uncertainty, and work related satisfaction. We used analysis of variance, including tests for linear trend where indicated, to test these associations. Because units of all scales are arbitrary, we categorized the continuous predictors into quartiles, because correlation or regression coefficients can be difficult to interpret, while quartiles are more intuitive. Furthermore, this approach also allows the reader to check whether the association is linear or not. Multivariate models were used to adjust for potential confounders, identify independent risk factor of higher cost-consciousness score, and compute adjusted means. All statistical tests were two-tailed, with a significance level of 0.05.

The distribution of the answers to most items was skewed toward the higher end of the response scale (Table 2). Almost all doctors (90%) agreed that trying to contain costs was their responsibility, and that they should take a more prominent role in limiting the use of unnecessary tests (92%). A majority of respondents (69%) disagreed with statements that doctors are too busy to worry about costs, and that the cost of a medication is only important if the patient has to pay for it out-of-pocket (88%). Only one out of four doctors (27%) thought that it is unfair to ask doctors to be cost-conscious and still keep the welfare of their patients foremost in their minds. Almost two-thirds (61%) agreed that there is currently too much emphasis on the costs of tests and procedures. The cost-consciousness score could be computed for 1176 respondents (99%); its mean was 16.8, SD 3.6, and quartiles 13.8, 16.8, and 19.2 when scaled between 0 and 24 (21). Its ditribution was close to normal.

Cost-consciousness was similar in men and women (Table 3), but younger doctors had slightly higher scores. Doctors in the public sector were in general more cost-conscious, in particular the senior staff across all age categories (results not shown). Internal medicine specialists and paediatricians had higher cost-consciousness scores, while doctors in surgical and technical specialties had the lowest (results not shown). Doctors who saw fewer patients per week and spend more time with each new patient were also more cost-conscious. Cost-consciousness scores were also higher for doctors who reported lower stress from uncertainty and higher work-related satisfaction (Table 3).

Multivariate analysis identified the following independent risk factors for higher cost-consciousness: type of practice (doctors in public sectors had higher scores), lower workload (number of patients per week), lower stress from uncertainty, and higher work-related satisfaction (Table 4). Other factors that were significant in the univariate analyses, such as specialty and self-reported time with a new patient, were not anymore significant after adjustment for type of practice, self-reported number of patient seen per week, stress from uncertainty and work related satisfaction. No sociodemographic characteristics were statistically associated with cost-consciousness in the multivariate analysis. We also explored the possible interactions between these factors, but none were significant.

The difference between the private and the public sectors may have several origins. It is possible that working in a public hospital raises the doctors' sensitivity to health care costs. Alternatively, the public sector may attract doctors who are more cost-conscious than those who choose the private sector. This issue could be resolved by following doctors prospectively, as they move from the public hospital into private practice.

Regardless of the setting, doctors who saw fewer patients per week were also more cost-conscious. In a fee-for-service system, a doctor's income is directly related to the number of patients seen. Doctors who see more patients per week may be more concerned about their income and less by costs incurred by society. However, the relationship was similar among salaried hospital doctors, who have no such financial incentive. It is possible that hospital doctors who see fewer patients may have academic or administrative roles that cause them to be more concerned about health care costs.

The association between cost-consciousness and higher tolerance for uncertainty is compatible with a previous report which linked higher anxiety due to uncertainty with higher health care expenditures in a Medicare health maintenance organisation [16]. Thus there is evidence that higher stress from uncertainty is associated with higher expenditures and, as we have found, lower cost-consciousness. From this, it follows that lower cost-consciousness and higher expenditures could be associated. To our knowledge, no study has yet studied this relationship.

The association with work-related satisfaction was substantial. This is of interest as work-related satisfaction correlates with several health outcomes, in particular with the quality of care [36, 37]. Although this association may be causal in either direction, we believe that work-related satisfaction may influence positively cost-consciousness: doctors who are happy about their work may think more about societal issues than unhappy doctors. Unlike other correlates of cost-consciousness, keeping doctors satisfied with their work is a modifiable factor. Nevertheless, whether improvements in doctor satisfaction translate into greater cost-consciousness remains to be seen.

The main strength of this study is its population-based sampling frame; the results are fairly representative of the attitudes of physicians in Geneva, Switzerland. The response rate of 59% raises the issue of selection bias, but is comparable to other surveys that depend on the cooperation of busy practitioners [2, 38, 39]. We could not assess response bias in depth, because the two medical organisations gave us only the adresses of their members, with no other socio-demographic information. The only caracteristic that we could track was sex, and its distribution was similar among those who responded and those who did not.

The main limitation of the study was its cross-sectional design, which precludes any formal conclusion about the causality of the associations between cost-consciousness and its correlates. Second, we were not able to obtain direct measures of doctor "costliness", such as the mean cost per patient, and thus cannot establish whether cost-consciousness influences actual costs of patient care. Third, we do not know to what extent respondents gave socially desirable answers about this rather sensitive topic. Finally, as the scales we used are not perfectly reliable, random misclassification may have caused the relationships we observed to appear weaker than they are in reality.