The prevalence of 'delayed or avoided physician visits' increased from 35.8% in spring 2004 to 46.8% about six months later, stayed high until spring 2006 and decreased to 35.7% in autumn 2006. Also, most subgroups of the respondents (e.g. defined by age and income) showed a very similar increase and decrease over the course of these six surveys. It can be concluded that it took some months for the reaction to this new regulation to fully develop, that about 21/2 years after its implementation its effect started to diminish, and that the social differences in this reaction remained rather stable in this period.
However, it is difficult to fully explain the observed time trend and especially the sharp decline between spring and autumn 2006, because avoiding or delaying physician visits is a very complex issue with many influencing factors. The question concerning the physician charge was identical in all surveys and no major political change took place in this time period. It has to be stressed, though, that the association of primary importance here (e.g. between income on one hand and delayed or avoided physician visits on the other) remains surprisingly stable in all six cross-sectional surveys (with different persons included in each survey). If possible, future studies should be based on a longitudinal design, as this allows for a much more precise assessment of time trends (and causal effects), of course.
Younger people were more likely to avoid or delay a physician visit than older. This is rather plausible, as usually they are healthier than older people and need to see a physician less urgently. The age gradient can still be seen after controlling for self assessed health and the presence of a chronic disease. This is probably due to the fact that the need to see a physician is not fully adjusted for by these two health variables. As expected, those who state that their health is poor or that they have a chronic disease delay or avoid a physician visit less often than those who are healthier (see Figure 5). Controlling for the other variables in the logistic regressions, though, the influence of these health indicators is rather small. It is significant only in the model focussing on the outcome 'avoided' (see Table 2, model 4). A third possibility to assess the presence of a chronic disease is presented by the variable 'maximum co-payment 1%'. Whereas about 49% of all participants state that they have a chronic disease, only about 14% state that they are exempt from the standard of 'maximum co-payment 2%' (see Table 1), clearly pointing to the fact that the indicator 'maximum 1%' indicates the presence of more severe diseases. This is probably the reason why this indicator for health shows the strongest association with the dependent variables.
Concerning the independent variable of primary importance here, i.e. income, the bivariate analyses clearly show that delaying or avoiding physician visits is reported most often in the lowest income group (see Figure 3). The association can also be seen in the multivariate analyses. In the subgroup of respondents having a chronic disease, for example, this reaction is reported in the lowest income group 2.45 times more often than in the highest income group. Also, a dose response association can be seen here (i.e. decreasing odds ratios with increasing income).
The fact that the Pseudo R2 values are rather small indicates that the variables included here are not sufficient to fully explain the variance of the outcome 'delayed/avoided'. We selected the independent factors after testing them for univariate significance and could of course only choose among those characteristics that had been included in the original study questionnaire. It is quite probable that a number of other aspects play a role in a patient's decision to consult a physician or not, namely his or her time budget, good or bad past experiences, travel distances to the physician's office, the physician-patient relationship or language barriers.
The results of the logistic regression can be used to calculate prognostic scores. Taking model 1 (Table 2), for example, the minimal risk score is obtained for a male person, who is more than 70 years old, earns more than 1300 € per month, has strong health awareness but poor self assessed health, has a chronic disease, pays only 1% for co-payments and has participated in the survey of spring 2004. His probability to delay or avoid a physician contact because of the practice charge is only 7.38%, whereas it is 81.62% for a young woman with the maximum risk profile. Thus, the odds ratio of a person at maximum risk (as compared with a person at minimum risk) reaches 55.72. These calculations are just intended to illustrate best and worst case situations. In reality, very few people belong to these extreme categories (in our study 8 old men and 5 young women).
The data used in this analysis are obtained by consecutive cross-sectional surveys. There is no follow up information per person over time, and therefore it is difficult to assess causality. It seems to be rather plausible, though, to assume that income has a causal effect on avoiding or delaying physician contacts due to the practice charge. Another problem is of primary concern here: The need to see a physician could only be assessed in a crude way, i.e. by taking into account measures of health and chronic disease. The final objective should be to analyse the health effects of delaying or avoiding physician visits, and this will only be possible with a better assessment of the need to see a physician, and with follow up information on health. Future studies should try to fill this gap. The present analysis can just point to the fact that negative health consequences of the new 'Praxisgebuehr' are probably most prevalent in the lowest income group (thus increasing health inequalities). Another limitation of the study is the fact that we could not assess if a person was exempt from the physician fee during the time period asked for in the survey (past three months). Most people accumulate their bills and ask for refund by the end of the year, but other may apply for exemption before the end of the year; these details are not asked for in the questionnaire, though.
Co-payments have been introduced in many industrial countries in order to minimize health insurance expenditures, to close budget gaps in the public health sector and to restrict moral hazard. As mentioned above, the concept of insured people 'over-using' insurance services is derived from insurance theory. M.V. Pauly hypothesised in 1968 that such 'excessive' demands could be depleted effectively by monetary hurdles such as co-payments . However, health care systems are very different from other markets regulated by supply and demand. Patients cannot choose treatments and medications like TV sets or beer brands, but have to rely on physician's decisions and recommendations. Moreover, medical services and products are rarely a matter of taste like luxury goods.
In particular, socio-economically disadvantaged people tend be less concerned about their own health , even if medical services are offered free of charge (such as preventive medical services in Germany), and co-payments can be an important additional financial hurdle. H. Reiners argues that the moral hazard concept is utterly misplaced in the public health sector, and that it is more plausible to assume that physicians are 'over-using' the system by inducing demand . The moral hazard argument has initiated highly controversial discussions in many countries, since co-payments and cost sharing schemes are frequently applied as a 'one-size-fits-it-all' tool in health politics, often disregarding the potential for jeopardizing health care especially for the poor. It is very difficult, of course, to determine a level of co-payment that discourages unnecessary utilization of services, and that does not discourage patients from seeking medical services they really need . A growing body of literature suggests that co-payments may adversely affect health outcomes . To date, the RAND-Study, conducted in California during the 1970s, is regarded as a fundamental investigation on moral hazard in health insurance . 5,809 US citizens were randomly assigned to 14 different health insurance contracts (with different co-payment modalities) and their consequent behaviour was documented for three to five years. One interesting result was that patients with higher co-payments abstained from necessary physician visits and had worse health outcomes in the end (e.g. concerning their teeth, blood pressure and eye sight). Furthermore, cost sharing had particularly negative effects on people with low income and shortened the lifetime of high risk patients . The negative effect of co-payments especially for socioeconomically disadvantaged groups could also be seen e.g. in studies from Israel [27, 28], South Korea , France [30, 31] and Denmark . In Austria and the Netherlands, some co-payments have been abolished after thorough evaluation, because they turned out to deter socio-economically disadvantaged patients from physician visits . A study in Canada documents that an increase of cost-sharing for prescribed drugs resulted in a decrease in essential medication among poor and elderly patients . A decrease in the utilization of life-sustaining drugs was also found in an international Cochrane review including 21 studies . Thus, there is increasing evidence of serious adverse effects that could ultimately lead to higher health care expenditures.
In a discussion paper on European strategies for tackling social inequities in health, the WHO states that denying access to effective health care is a denial of human rights. Nevertheless, the existence of inequities in access to health care can even be found in the most advanced welfare systems in Europe . A health care system is needed that guarantees basic, affordable health care coverage for all citizens without discriminating specific socio-economic groups. Co-payments may pose a major threat to this principle of solidarity, they could lead to increasing health inequalities, and also to higher health care cost for disadvantaged population groups in the long run. It would be important to assess these potential problems in more detail, and in advance (e.g. by conducting health inequality impact assessments before a reform such as the 'Praxisgebuehr' is implemented). The financial resources are limited, of course, and in some circumstances co-payments could be helpful for adjusting the provision of health care services to health care needs. They should not be an instrument for increasing health inequalities, though.