Hyperprolactinemia is the most common pituitary hormonal disorder [10] although it is still considered a rare condition. With the current incidence estimate of 5 per 100.000 person years, we could expect approximately 792 individuals to develop hyperprolactinemia in the study period. However, we found more than 7 times that number of individuals with hyperprolactinemia and more than 100 times that number of individuals have had a measurement of prolactin performed.

The current recommendations form the Endocrine Society is to measure prolactin if hyperprolactinemia is suspected [4], but indications other than suspected hyperprolactinemia may prompt a prolactin measurement, e.g. routine follow-up of patients with pituitary disorders including pituitary mass lesions and hypopituitarism. A common complaint presented in general practice is menstrual disturbances which has a prevalence of 20.7% in a group of fertile young women [11]. The most common cause of oligomenorrhea and secondary amenorrhea is polycystic ovarian syndrome (PCOS) [12] and another frequent cause is premature ovarian insufficiency (POI) [13]. Endocrine Society and the European Society of Human Reproduction and Embryology, recommends measuring prolactin in females with oligo- or amenorrhea or when suspecting PCOS or POI [14].

Moreover, prolactin may be more likely to be measured in individuals receiving treatment for a psychiatric disorder [15, 16] as increased prolactin is a well-known adverse effect of several anti-psychiatric and anti-depressive drugs [17, 18]. Several other types of medication e.g. antiemetics also have the potential to elevate prolactin levels. However, the use of these medications is probably less likely to warrant a measurement of prolactin as the use of medication for psychiatric disorders.

We found a surprisingly high number of prolactin measurements in our population of whom the majority (68.1%) only had a single measurement, which accounted for 36.5% of all the measurements. Thus, these individuals probably had symptoms indicating hyperprolactinemia, which was ruled out by a normal blood test, as recommended by current guidelines [4].

A large proportion of those tested exhibited repeatedly normal prolactin levels, accounting for 41.1% of all the measurements in our study. This is a considerable number of tests, which suggests over usage that may relate to either prolactin measurement being part of a routine work up program or the result of so-called defensive medicine, which is prevalent according to both Danish and international studies [19, 20].

On the other hand, a surprisingly large number of persons, 35.7%, with elevated prolactin never had a follow-up measurement.

To our knowledge, no previous studies have prospectively detailed the pattern of prolactin measurements in a large population-wide cohort. It was expected that a large proportion of the tests were ordered by a general practitioner, since they are gate keepers and provide primary care also for premenopausal women with menstrual irregularities, which frequently triggers prolactin assessment [21]. The reason for the large number of repeated tests performed by the general practitioners, however, is unexplained and merits auditing.

The cost of prolactin tests in the study amounts to approximately 590.000€. Given the fact, that the Danish healthcare system is publicly funded and health expenditure amounts to 10.8% of the gross domestic product (GDP) [22], cost-effectiveness is a constant concern. Our price estimate is only for measuring prolactin per se, and the total cost of ordering a prolactin measurement (and likely a number of other tests at the same time) is probably much higher when accounting for e.g. the derived expenses for follow up visits, MRI, referrals to specialized medical departments. Furthermore, in many individuals, the detection of hyperprolactinemia was not followed by prolactin lowering therapy. There may be several reasons not to initiate treatment in persons with elevated prolactin e.g. lack of knowledge about the disorder in general practice, lack of symptoms in eugonadal individuals or individuals where the cause of hyperprolactinemia is suspected to be caused by treatment for a psychiatric disorder or other medications like antiemetics. Lack of treatment with DA could also be a patient preference or be owing to a different treatment strategy e.g. estrogen supplements in females with no fertility wish. If treatment is not initiated, regardless of the reason, the relevance and benefit of repeated measurements in persons with established hyperprolactinemia is questionable.

We found a surprisingly high number of patients with elevated prolactin in the study (20.3%), compared to what has previously been reported [5, 6]. However, previous data are derived from individuals receiving treatment with dopamine agonists, which likely underestimated the number of individuals with hyperprolactinemia. We chose to include the category “elevated prolactin” in our study to report the number of high measurements. Prolactin measurement is highly sensitive and an elevated prolactin level is not necessarily a pathological finding. Prolactin can be transitorily elevated for a number of reasons like medication, physical activity and sexual intercourse. The transitory nature can be seen by the fact that more than 80% of persons with elevated prolactin only had a single elevated measurement. This is also the reason we used the more stringent categorization for hyperprolactinemia requiring at least two elevated prolactin measurements.

Our study finds a likely overutilization of prolactin measurements. This may be owing to a combination of a generous screening strategy and the lack of knowledge of the effects of having mildly elevated levels of prolactin without relevant symptoms. Further studies evaluating the effects of elevated prolactin in asymptomatic individuals, on e.g. Fertility, morbidity and socioeconomics, are needed in order to gain more knowledge. This will enable the possibility for better recommendations regarding the measurement of prolactin, including when not to measure prolactin, when to initiate treatment and how to monitor these individuals.

Our study has several strengths. It includes data on a large population, with many measurements. Furthermore, the use of registry data enabled us to link use of medication and information of MRI to the prolactin data. The data in our study are real life and thus representative of current clinical practice.

Our study also has some weaknesses. The use of registry data has an inherent risk of error due to poor reporting. This risk is low in this study since the BI-portal automatically transfers data from the laboratory, and thus is not dependent on reporting. However, the fact that we set a fixed study period may result in a slight inaccuracy in our results due to the timing of measurement and initiation of treatment. For instance, more people (n = 804) received treatment than were classified as having hyperprolactinemia (n = 736) which may suggest that a few individuals started treatment before the start of our study or have moved to Central Denmark Region on dopamine treatment. Likewise, some individuals may have had an elevated prolactin measurement at the end of our study period but no record of treatment leading to misclassification of these individuals. In the same way our chosen cutoff of DA treatment for less than 30 days might have resulted in misclassification of a few individuals who may have started treatment within a month of the end of the sampling period. However, due to the long study period, we consider this a negligible risk of misclassification.

We included only the use of medication used to treat psychiatric disorders. It is well known that several other types of medication can affect prolactin levels. The lack of data on all medications used by the persons in the study is a limitation for the understanding of our findings. However, we do not believe that additional information on medicine use would alter our findings in any significant way.

Using data from registries entails a lack of granulated clinical data. Thus, we have no data on possible symptoms or the clinical reasoning for ordering blood tests. This limitation hampers a deeper interpretation of our data in a clinically significantly way.

The number of prolactin measurements was higher than anticipated, but the number of elevated prolactin levels was also quite high. Still, the number of patients in whom an elevated prolactin level led to dopamine agonist treatment remained relatively low.

Further studies examining the use of prolactin in a clinical setting as well as the clinical outcome of elevated prolactin levels are needed in order to provide future guidelines for the rationale of measuring prolactin.