METHODS

In March 1996 questionnaires were mailed to the parents of 15 255 children aged 2 to 17 (around 3000 in each country), who were randomly selected from the national population registries. Two reminders were sent to non-responders, except in Norway where only one remainder was permitted. The questionnaire was translated from Swedish into Danish, Finnish, Icelandic and Norwegian and backtranslated by bilingual persons.

The study design and the response rate have been described in detail elsewhere.¹¹ The response rate was highest in Finland (79.5%), about 69% in Denmark, Sweden and Iceland, and lowest in Norway (64.5%). Non-response analysis was done to a different degree in the five countries. In most cases it was done by the statistical bureaus in each country. There was hardly any difference in response rate according to the sex of the child in any of the countries and little difference between age categories. Families with low education, working class and one parent families were over-represented among non-responders. Little difference was found according to urbanisation grade. In the country with the largest non-response rate, Norway, the largest difference in response rate was for education of the mother, with 80% response rate in the highest educational category but only 50% in the lowest category. This difference according to education was much smaller in Sweden, for example.

Use of health care was measured by asking whether the parents had consulted a general practitioner (GP), a specialist or had a phone call to any doctor for their children during the previous three months, and whether the child had been admitted to a hospital during the previous year. Visits for health check ups were not included.

Indicators of SES used in this study were the educational levels of the father and the mother, and the family income. Educational level was measured as the highest level of completed education. Four levels were distinguished: university/college level, upper secondary school education (>12 years), lower secondary school education (10–12 years) and primary school only (<10 years). Family income was measured as the disposable monthly income of the family. It comprised salary, pension, income from own firm and benefits and allowances (for example, child benefits, maintenance benefits, housing benefits, and social benefits). Income was measured as a continuous variable. In each country it was divided into four categories, from the highest to the lowest, with roughly equal number of children in each group.

Three models were used to analyse the association between socioeconomic variables and each kind of health service used. Model I adjusted for age (three-year age groups), sex and urbanisation grade. Urbanisation grade was measured as living in a town of more than 100 000 inhabitants, a town or village with a population between 3000 and 100 inhabitants, or living in a rural place with population less than 3000 inhabitants.

Model II additionally adjusted for health status in order to take into account socioeconomic differences in need for health services. Two health indicators were used: having at least one chronic condition, and experiencing at least one moderate or severe symptom at least every other week. The list of chronic conditions included diabetes, visual impairment, hearing impairment, speech defect, psychiatric disorder, epilepsy, gastric disorder, asthma, allergies, eczema, physical disablement, overweight, and hyperactivity. The symptoms included were stomach pain, headache, back pain, disorders of sleep, dizziness, and loss of appetite. The lists of chronic diseases and symptoms were compiled by experienced paediatricians in the Nordic countries, but they have not been tested for validity or reliability.

In model III we additionally adjusted each socioeconomic factor used in these tables for all other socioeconomic factors, in order to determine the independent contribution of this factor to the socioeconomic differences found.

RESULTS

Table 1 shows the proportion of the children who had used different services during the preceding three months, and the proportion who had been in hospital the previous year, by age group, sex, country, and degree of urbanisation. The youngest of the three age groups used most services, while there was little difference between the other age groups. Gender differences were small. The physician services, especially specialist services, were used less in rural areas. Icelandic children used most services and Swedish children fewest, surpassed only by Norwegian children—for specialist services.

Table 2 shows in the first column the proportion of children with parents in each socioeconomic group in the Nordic countries as a whole. The remaining part of the table shows the socioeconomic differences in children's use of health services. Socioeconomic differences were more distinct if not adjusted for urbanisation grade (results not shown), obviously because people in higher socioeconomic classes more often live in urban areas, where use is higher (table 1).

In the Nordic countries as a whole the use of GP services was similar in different socioeconomic groups. The exception was a lesser use of GP services by children who have mothers with only primary school education, but the difference is not statistically significant.

There were considerable socioeconomic differences in the use of phone calls to doctors according to educational level of the parents. The lower the education of the parents the lower was the use of telephone consultations with doctors. This association was stronger in the case of mother's education, as seen in model III where other factors have been adjusted for. There was no association between phone calls and income of the family.

Children who had parents with lower education and income used less specialist services than children did in higher socioeconomic groups. The difference between the highest and the lowest category was statistically significant for all three socioeconomic variables (model II). Here again the mother's education stands out as the most important socioeconomic factor after adjusting for the other two socioeconomic variables (model III).

Hospital services for children showed quite an opposite picture, with higher odds ratios in the lower socioeconomic groups in model I and II. The results were statistically significant for the lower educational categories of the parents and especially for the mother.

Table 3 shows the socioeconomic differences in the use of physician services in each of the Nordic countries after adjustment for health, age, sex, and urbanisation grade. In most countries the same picture was seen as the one discussed above, but with some exceptions. In Denmark the GP services were less used by children with parents with lower education and income. In Finland GP services were used significantly more often by the lowest income group. Phone calls were generally used less by the lower classes in all countries except in Sweden. Specialist services were also generally less used by the lower classes in all countries, but the difference was less marked in Sweden and Iceland. Hospital services were more used by the lower socioeconomic groups in all countries with the exception of the lowest income groups in Iceland and Sweden. The positive association of hospital use with educational level of the parents was most pronounced in Denmark.

DISCUSSION

The main findings of this paper are that there were little differences according to SES in the use of GP services in the Nordic countries as a whole, that phone calls to doctors and use of specialist services were significantly less used by lower class children, but lower class children were more often admitted to hospital. The use of services had stronger associations with the parent's education, especially the mother's education, than with the family income. Similar associations were observed between for each country separately with a few exceptions. In general, differential use was more marked in Denmark, Finland and Norway than in Sweden and Iceland.

The question may be raised whether the results are biased attributable to non-response. It was observed that non-response was higher in lower socioeconomic groups (see Methods). This higher non-response may have underestimated the use of the services by children in lower groups (and of inequalities in their health care use) if non-response among parents is related to higher health care use of their children. Even though there might perhaps be such an association, there is no reason to expect this relation to be strong. Therefore we would not expect a substantial effect of non-response on the observed inequalities in health care use.

The accuracy of parent reports is assumed to depend on recency and seriousness of an event. Pless and Pless¹² studied the accuracy of parent recall compared with paediatrician records and found it to be unrelated to demographics such as respondent's age and education. Other studies on parent recall according to SES in surveys of health care use of children support this finding.^13,14

Adjusting for chronic diseases and symptoms had rather small effects. Children in low socioeconomic groups in the Nordic countries do have more chronic health problems than those in higher socioeconomic groups.¹⁵ Chronic diseases are not as common among children as they are among adults and physicians see children more often for acute diseases. The expected need of the children can only be partially controlled for in this study, as no information on recent acute diseases was available. It may be expected that, if control could have been made for more accurate measures of need, we would have observed larger inequities in health care utilisation.

These findings, especially with regard to the specialist services, give reasons to ask whether the co-payment in the Nordic countries¹⁶ acts as a barrier for the use of the services by parents in the lower income groups. In Denmark both GP and specialist services were free at the point of use in 1996. In Finland there was no charge for persons under 15 at the health care centres. In Norway treatment for children under age 7 was free of charge but other children paid $18 for a GP and $25 for a specialist consultation. Icelandic children paid a reduced fee of $3 for a GP consultation and $7 for a specialist. The highest co-payment was in Sweden: the local authorities set the charges and they varied between $10 and $20 for a GP and between $25 and $45 for a specialist for children at any age.

In Sweden there is concern about rising co-payment in outpatient care. The Level of Living Survey in 1993–4, conducted on a regular basis in Sweden, showed for the first time in many years that there was a statistically increasing difference between educational groups in the use of outpatient health care for adults, adjusted for health status. This was referred to as “cracks in the welfare wall”.¹⁷ However, no differences were found in the use of health services for children between income groups. Other studies¹⁸ indicate that people refrained from consulting a doctor because of cost and, as could be expected, this was more common among people in low income groups. These results have been challenged because of the risk of tactic answers and lack of answer alternatives regarding reasons for not visiting a doctor.¹⁹

Denmark is of special interest as there is no co-payment for the services and as only in Denmark the GP acted as gatekeepers to the specialist services, even though people in all of these countries were expected to contact their GP in the first hand. Denmark showed larger differences than Sweden in use of GP services according to family income, but both countries had similar low levels of differential use of specialist services according to family income. If co-payment were a barrier we would expect larger inequalities in the countries with higher co-payment, especially when comparing Denmark with Sweden on the opposite ends of the co-payment spectrum and with the same response rate. No such relation was found.

In general there were differences in the use of specialists according to the educational levels of the parents, especially the mother's education. This indicates that better educated parents are more able to go directly to a specialist, to by pass the gatekeeper function of the GP or negotiate for referral to a specialist. This difference was not smaller in Denmark with its referral system to specialist care than in most of the other countries. Mother's education may determine health care use in several ways, including the suggestion of earlier research that the more educated mother is more likely to interpret her own child's symptoms as abnormal, while still being able to discriminate their relative severity.²⁰ If the same applies to this study, underutilisation of health services by low educated mothers is a more probable explanation than overuse by the more educated.

CONCLUSION

The specialist services and use of telephone services for children in the Nordic countries do not meet the criteria of equal use for equal need whereas the GP services and hospital services do to some extent. The comparison between countries shows no relation between different levels of co-payment and differential use of physician services by income groups. Low education of parents, especially the mother, is a more important determinant of health care utilisation for children than the financial situation of the families in the Nordic countries. The finding that telephone services for children in lower classes, with very low direct costs, are not used to the same degree for children of better educated parents also supports this.