Statement of principal findings
This study did not find a significant difference between teams with an NP and teams with only GPs with regard to X-rays, drug prescriptions and referrals to the ED. Moreover, the production per hour and the cost per consultation for the team with an NP were not different from teams with only GPs.
In the experimental team, NP care was found to be associated with significantly fewer drug prescriptions and fewer ED referrals than care delivered by GPs. NPs were shown to have a lower production per hour than GPs. The cost per consultation with an NP was lower than with a GP.
Strengths and weaknesses
A strength of the current study is its large patient sample and a long follow-up period, but limitations include the single-centre character of the study and the low number of nurse practitioners involved. Moreover, we had a relatively large number of missing ICPC codes. There appeared to be only a few GPs who repeatedly did not report ICPC codes, which means the bias is related to the GP and not the ICPC diagnosis or day. This is supported by the fact that the ICPC codes in our study are comparable to those of other out-of-hours services in Western countries [
18]. Therefore, we don’t suspect that the missing ICPC codes will cause any bias to our outcomes.
It should be noted that the current study shows the effect of NPs within a GPC. Although many countries have organized out-of-hours care in large-scale organizations in previous years, the various types of health-care systems influence the generalizability of the research findings [
15]. Moreover, the education and deployment of NPs differs between, and even within, countries and health-care systems. In the Netherlands, as in most countries, NPs providing care are always working as part of primary care teams alongside GPs [
21,
24]. Our results can therefore not be generalized to other models of care in which NPs are working in teams without GPs [
25]. Moreover, in the current study the NPs were primarily responsible for treating minor ailments. The complexity of tasks can differ between regions and countries.
In the current study, NPs with no experience working at the GPC at the start of the study were compared with GPs who had on average 7.3 years of experience at the GPC. This may have influenced resource use or production per hour. A strength of the current study is the fact that researchers did not change patient allocation, which gives an accurate representation of the daily practice and related cost estimates.
We only included costs relevant from the GPCs’ viewpoint (tariff per hour, production per hour) and direct health-care costs relevant from health insurance companies’ viewpoint (X-rays, drug prescriptions and referrals to the ED). This implies that it is not possible to draw conclusions on whether the deployment of NPs is cost saving from a societal viewpoint. Therefore, other factors, such as the difference in costs of training, rates of sick leave, patient follow-up after a GPC visit or after ED referral, et cetera, should have been included [
23,
26].
Comparisons with other studies
Meta-analyses based on research conducted in daytime primary care did not show differences between nurses and GPs in terms of prescriptions, diagnostic test orders and referrals [
10]. Although, in line with these meta-analyses, we did not find differences at team level, our secondary analysis in the experimental team showed a difference between GPs and NPs in terms of drug prescriptions and referrals to the ED. We cannot determine whether this difference in resource use is an overuse of medication or referrals by GPs, or an underuse by NPs. There is no capacity to examine how clinical outcomes would differ from the likely outcomes if patient care was provided by the other care provider [
27]. Inappropriate referrals and prescriptions may further increase health-care costs and unnecessary treatments in the hospital. Based on reviews of research, we do not expect an underuse by NPs since patient outcomes in primary care were found to be at least equivalent for NPs and GPs [
12,
14]. Moreover, research on the ED and hospital care shows that the diagnostic accuracy of NPs is comparable to that of doctors [
28,
29].
We found a lower production per hour for NPs than for GPs. However, it was not possible to adjust this outcome for case mix. This makes comparison between GPs and NPs difficult since they treat different patients. However, we expect the number of consultations per hour to be a reliable measure. This is supported by the fact that our outcomes are comparable to results from meta-analyses on consultation times [
10]. Besides treating different patients, lower production per hour can also be associated with less experience [
30]. Although NPs had at least five years of experience in primary or elderly care, none of them had any experience in out-of-hours primary care at the start of the study. Other possible explanations for longer consultations include a higher use of protocols [
10], and a more holistic approach and greater provision of information by NPs than by GPs [
31]. In addition, the provision of more health education and information by NPs may result in fewer prescriptions [
32].
Based on previous research, we expected NP care to be cost saving due to a lower salary for NPs than for GPs [
33]. However, in line with another study, lower production per hour appeared to lessen the influence of salary differences on consultation costs [
34]. Another reason for the small influence of salary costs on overall costs is the small difference in tariff between the GPs and NPs during out-of-hours care. This is because the GPs receive financial compensation for out-of-hours care based on the total tariff for providing care to their patients 24/7. This means that the GPs receive a fixed tariff, whereas the tariff per hour for NPs was based on their gross salary including social security contributions and premium pay. The differences in tariff per hour would have been bigger in cases where the care providers were employed by the GPC in the same way. For example, the difference in gross salary of a GP employed by another GP and the NPs in our study is approximately 60 % [
35]. In another Dutch study in daytime primary care, the salary of an NP appeared to be less than half of that of a GP. As a consequence, in that study, cost differences were mainly caused by the difference in salary [
36]. It is expected that bigger differences in salary will result in more cost savings when GPs are substituted with NPs.
The current study shows that the differences in referral rates to the ED strongly influenced consultation costs. The fewer referrals by NPs resulted therefore in lower mean costs of care provided by NPs than by GPs. It is difficult to compare these findings with previous research due to conflicting results on the effect of substituting GPs with NPs in primary care on the cost of health care. Moreover, due to heterogeneous outcome reporting and the small number of studies they are hard to interpret. However, in general, NP care seems to be associated with lower or equal health-care costs per consultation [
6,
12]. Only one study found increased costs associated with NP care. These results were based on two factors that we did not measure: time spent by GPs on supervising and number of return visits [
34]. The time spent on supervising in the current study was, however, relatively low. The NPs consulted a GP in only 7.1 % of all consultations. Only 0.2 % of the patients were taken over by the GP; the other consultations were completed by the NP. Consultations between the NP and GP are considered part of daily practice and comparable to consultations GPs have with other GPs Therefore, we do not expect this to bias our outcomes.
Study implications
The current study shows no differences in resource use and direct health-care costs between teams with an NP and teams with GPs only. Therefore we conclude that during out-of-hours, involvement of NPs in multidisciplinary teams can increase capacity without increasing resource utilization.
Our results show that using NPs as substitutes for GPs in out-of-hours care is a feasible solution for decreasing GPs’ workload or increasing service capacity. It should be noted that tasks at GPCs are limited to providing acute care and do not use NPs’ competences to the full. Tasks such as preventive projects, psycho‐social home visits, providing ongoing training for staff and developing protocols are only performed during the daytime. In countries where GPs deliver 24/7 care, the implementation of NPs in primary care will only succeed when they (just like GPs) provide care 24/7.
With the need for extra workforce in primary care, our data suggests that substitution by NPs can be considered an solution economical equal to the care delivered by GPs. However, because we only included one GPC, and only measured direct costs, results should be interpreted with caution. Economic evidence on which to make judgments on future out-of-hours care is far more complicated [
37]. Other costs from a societal perspective such as training cost and unemployment rates of physicians in hospital care have to be taken into account. This implies that decisions on the substitution of GPs by NPs in out-of-hours primary care should not only depend on costs, but on other factors such as a view on professional roles, responsibilities, and quality and safety of care [
34].
As this study showed a significant difference in cost per consultation in favour of NPs, it may be possible that deploying more NPs in a team with GPs is more cost saving. Future research is needed to indicate an optimal balance in which teams with NPs and GPs provide the most efficient care for patients in out-of-hours primary care.