Repeated Measurements of Vitamin D: Common and Very Unlikely to Indicate Deficiency

ABSTRACT

Objective

Vitamin D testing uses considerable resources and has increased recently. We investigated repeating testing in Auckland, New Zealand.

Design, Measurement and Patients

We analysed repeated 25‐hydroxyvitamin D (25OHD) tests in three databases from the Auckland region from 2005 to 2024 (362,618 tests). We assessed annual numbers, time between tests, characteristics associated with repeat tests, and relationships between requestors and repeated tests.

Results

Twenty‐four percent of individuals had ≥ 2 25OHD tests. 34% of all 25OHD tests were repeats. A disproportionately high number of repeat tests were done in a small number of individuals: 57% had one repeated test (i.e., two tests total) but 5.3% had ≥ 6 and 1.4% had ≥ 11 repeated tests. Fifty eight percentage of repeat tests were done within 1 y and 77% within 2 y. 85% of people with repeated tests had a first 25OHD > 25 nmol/L, but in only 9.8% a repeated 25OHD was ≤ 25 nmol/L. Most repeat tests (77%) when the first 25OHD was > 75 nmol/L were > 75 nmol/L A small number of high‐volume requestors ordered most tests: 3/7637 ordered > 5000 tests, collectively 7% (21,092) of the total, and 98 ordered > 500 tests (collectively 44%). The most frequent requestors were primary care doctors and bariatric surgeons. Simple restrictions based on the result and time since previous 25OHD could reduce repeated tests by > 60%–90% with only 4%–5% of 'avoided' tests ≤ 25 nmol/L.

Conclusion

Repeated vitamin D testing is common, and most results are not low. Restricting test availability and targeting those at risk of clinical complications of vitamin D deficiency are likely to reduce unnecessary tests.

1. Introduction

Vitamin D testing consumes considerable health care resources. For example, large increases in the demand for testing in the UK caused increased workload, large backlogs in testing, and increased costs [1]. The annual cost of testing in the UK was estimated to have increased from about £1 m–£17 m between 2001 and 2018 [2]. Similar costs and consequences of increased vitamin D testing have been reported in Australia and Canada [3, 4]. We reported similar findings in Auckland, New Zealand, with increases in vitamin D testing [5] that were largely driven by a few high volume requestors [6], despite messaging and education about cost and utility on all 25‐hydroxyvitamin D (25OHD) result reports. As a consequence of those increases, restrictions were introduced for vitamin D testing in Auckland in 2012, which immediately decreased the number of tests requested [5] but did not increase the proportion of tests with low results [7]. In Australia, changes to testing guidance designed to restrict testing (but without formal restrictions) led to a short‐term reduction in requests that was not sustained and requests had exceeded pre‐guidance levels within 2 years [8]. Paradoxically, after the guidance was introduced, there were greater proportions of test results > 50 nmol/L and smaller proportions with results < 25 nmol/L [8]. In Ontario, Canada, testing increased 10‐fold between 2006 and 2010, leading to the implementation of restrictions which immediately reduced testing by 82%, but testing steadily increased and returned to the pre‐restriction peak after about 13 years [4]. Given the consequences for costs and workload of unnecessary testing, we investigated trends in testing and the characteristics of repeated 25‐hydroxyvitamin D (25OHD) tests in Auckland, New Zealand.

2. Methods

We used three overlapping datasets in these analyses. The Auckland Regional dataset contained deidentified data from the Auckland regional biochemistry database for all measurements of 25OHD between 1/1/2009 and 15/6/2020. These measurements were carried out in different laboratories and used different 25OHD assays, including Roche, Diasorin, Siemens, and Abbott immunoassay platforms. For each 25OHD measurement, we identified any relevant contemporaneous biochemical test results (serum calcium, phosphate, alkaline phosphatase, parathyroid hormone), defined as the closest result up to 6 months before or after the 25OHD measurement. The other two datasets were from Auckland LabPlus (public hospital and community‐based requests 1/1/2005–31/12/2024, using Diasorin and Roche assays), and Awanui labs (community‐based requests using Siemens Centaur and Abbott Architect platforms), and contained all 25OHD results and requestor details. Both laboratories had continuous ISO15189 accreditation and complied with performance recommendations on external EQA schemes (RCPA and/or DEQAS) throughout. The study was approved by the Northern B Health and Disability Ethics Committee (2025 EXP 21400).

The three datasets were merged and duplicate records were identified and removed. Requestor primary care or specialist information was available for the majority of requestors, but where this was unavailable for the 200 most frequent requestors, we sourced the specialty manually. Of 7637 individual requestors, specialist category was present for 49% of requestors who collectively ordered 88% of the tests. We assessed the number of single and repeated 25OHD measurements over time, the time between measurements, and the characteristics associated with repeated measurements in the complete dataset. In the subset of records with requestor details, we assessed the relationships between requestors and repeated 25OHD tests. Descriptive statistics (mean, SD, or proportions) are presented. All analyses were conducted with R software packages (R 4.4.2, 2024, R Foundation for Statistical Computing, Vienna, Austria).

3. Results

Summary demographic and 25OHD measurement characteristics from the three datasets are shown in Supporting Information S1: Table 1. The average age of people having 25OHD tests was 41 y, and 67% were female.

3.1. Repeated Vitamin D Measurements

Figure 1 shows the number of tests per year, categorised by age and the number of tests per patient (single or multiple tests). In 2012, restrictions were introduced which limited testing to patients at high risk of or being investigated for osteomalacia; and children < 16 years (any requestor), as well as all requests from endocrinologists, gastroenterologists, bariatric surgeons, rheumatologists and nephrologists. The restrictions led to a reduction in testing in adults. Total tests dropped by 82% from 2011 to 2014, by 90% in adults and 73% in children.

Figure 1.

Number of 25OHD measurements by year and by age. In 2012 restrictions were introduced for adults. The number above each bar is the total number of measurements per year, and the percentage is the number of individuals in each year with a single 25OHD measurement during the 20 year time‐period.

Since 2014 the number of tests has increased again by more than fivefold (6.2‐fold increase in adults, 3.7‐fold increase in children). The proportion of single tests prior to restrictions was 49%–57%, dropping slightly post‐restrictions to 43%–54%. After the restrictions were introduced for adults, testing in paediatric patients remained steady until 2020 and consequently made up a much greater proportion of requests (9%–12% pre‐restrictions, 29%–36% 2012–2020, post‐restrictions). Subsequently, the proportion of paediatric patients steadily dropped, reaching 24% in 2024 as the number of adult requests progressively increased.

One hundred eighty‐one thousand nine hundred twenty‐seven (76%) individuals had a single 25OHD measurement. 58,129 (24%) individuals had multiple 25OHD tests: in total they had 180,691 tests, or 50% of the total number of tests. They had 122,562 repeated tests, representing 34% of the total test numbers. The median number of repeated tests/person was two (i.e., three tests in total), with the mean 3 and the maximum 127. Table 1 shows demographic and biochemical results associated with each 25OHD test by whether the test was single or multiple and, for multiple tests, whether it was the baseline or a repeated test. There were no clinically relevant differences in these variables between the single test group and the multiple test group at baseline. The proportions of tests by month across the three groups and throughout the year were similar (data not shown).

Table 1.

Characteristics associated with single, multiple or any 25OHD measurement.

			Multiple
Variable	All	Single	Baseline	Repeated
N	362,618	181,927	58,129	122,562
Age (y)	41.4 (22.5)	41.4 (22.8)	40.2 (21.9)	42.0 (22.2)
< 18	71,615 (20)	35,224 (19)	11,521 (20)	24,870 (20)
18+	290,962 (80)	146,672 (81)	46,603 (80)	97,687 (80)
Gender
Female	241,751 (67)	114,665 (63)	40,733 (70)	86,353 (71)
25OHD (nmol/L)	64.3 (34.9)	60.4 (32.4)	56.4 (32.4)	74.1 (37.4)
≤ 25	37,881 (10)	22,098 (12)	8898 (15)	6885 (6)
≤ 50	137,355 (38)	76,227 (42)	28,380 (49)	32,748 (27)
Total serum calcium (mmol/L)	2.36 (0.17)	2.37 (0.16)	2.36 (0.19)	2.34 (0.16)
Associated calciuma	77,976 (55)	37,350 (53)	11,724 (58)	28,902 (58)
Calcium low	885 (1)	377 (1)	187 (2)	321 (1)
ALP (U/L)	117.9 (150.1)	111 (1455)	127 (162)	123 (152)
Median (IQR)	80 (62–117)	80 (63–113)	82 (63–125)	79 (61–120)
Associated ALPa	113,343 (81)	55,469 (79)	16,131 (80)	41,743 (83)
ALP high	4949 (4)	1922 (3)	767 (5)	2260 (5)
PTH (pmol/L)	9.2 (15.8)	9.6 (15.7)	8.8 (14.7)	9.0 (16.4)
Median (IQR)	4.9 (3.3–8.3)	4.8 (3.2–8.4)	5.0 (3.5–8.0)	4.9 (3.4–8.3)
Associated PTHa	17,608 (13)	6957 (10)	3399 (17)	7252 (14)
PTH high	5101 (29)	2031 (29)	971 (29)	2099 (29)
Phosphate (mmol/L)	1.24 (0.34)	1.22 (0.34)	1.27 (0.38)	1.25 (0.33)
Associated phosphatea	47,437 (34)	20,936 (30)	7367 (37)	19,134 (38)
Phosphate low	3335 (7)	1438 (7)	590 (8)	1307 (7)

Note: Data are mean (SD) or n (%) except where stated. Single refers to individuals with one measurement; multiple baseline to the first measurement of individuals with multiple measurements, and multiple repeated to any subsequent measurements.

^a Associated [n (%)] means a contemporaneous measurement within 6 months before or after the 25OHD test from the Auckland Regional dataset, and the percentage is the proportion of measurements in that dataset with a test result. For example, 77976 25OHD measurements had an associated calcium measurement, representing 55% of 25OHD measurements. For this analysis, a 25OHD below the lower limit was imputed as 8 nmol/L and > 250 nmol/L as 250 nmol/L. All comparisons of the single and multiple baseline groups were statistically significant (all p < 0.02, 15/20 p < 0.001), except for PTH high (p = 0.52).

Figure 2A shows the distribution of the frequency of the 122,562 repeated 25OHD measurements. 57% of people with multiple tests had a single repeated test (two total tests), but 5.3% had at least six repeated measurements (7+ in total), and 1.4% had at least 11 repeated measurements (12+ in total). Seven patients had > 50 repeated tests over 20 y (range per patient: 52–127 repeated measurements, range of period of measurements per patient 5–15 years, range of measurements/year per patient 3.7–12.2).

Figure 2.

Panel A: the frequency of repeated measurements, for example 33,096 individuals had one repeat measurement (two total) while seven had ≥ 51 repeat measurements; Panel B: time between consecutive measurements, for example 5.8% were repeated on the same day, and 1.6% within 7 days. Panels C–E: distribution of 25OHD for single or multiple baseline and repeated measurements. In A and B, the number and percentage above the bars are for individuals for that category. In C–E, the black bars show measurements below or above the relevant limit of the 25OHD assay.

Figure 2B shows the time between consecutive measurements. 37%, 58%, and 77% had their next test within 6 m, 1 y and 2 y, respectively, of their previous test, and in only 4.5% were the consecutive tests at least 5 y apart. 56% of individuals with at least three tests during follow‐up had an average of > 1 test/year between their first and last test, and 79% had an average of at least one test over 2 year.

Figure 2C–E show the distribution of 25OHD results in people with a single measurement and the baseline and repeated 25OHD results in people with multiple tests. Compared to the distribution of single results, in people with multiple tests the baseline distribution is skewed slightly to the left (lower 25OHD) and the repeated distribution to the right (higher 25OHD). 85% of individuals having repeated measurements had a baseline 25OHD > 25 nmol/L, and only 9.8% had at least one repeated measurement ≤ 25. Of the total number of repeated tests, only 6% were ≤ 25 nmol/L (see also Table 1).

Supporting Information S1: Table 2 shows the proportions of individuals with repeated 25OHD measurements for different 25OHD ranges by initial 25OHD. Four thousand seven hundred fifty‐four people with an initial 25OHD > 100 nmol/L had 6600 repeated measurements, of which only 78 (1%) were ≤ 25 nmol/l and only 8% ≤ 50 nmol/L, while 73% were > 75 nmol/L and 42% > 100 nmol/L. Similarly, when the baseline 25OHD was > 50 nmol/L or > 75 nmol/L, only 4% and 2% respectively of repeated 25OHD were ≤ 25 nmol/L, and 26% and 13% respectively were ≤ 50 nmol/L.

3.2. Requestors

Information on test requestor was available for 86% of test results with 7637 individual requestors (313,050 tests). Figure 3 shows the number of tests ordered by individual requestors, and the results of those tests. Over the 20‐year period, 26% of requestors ordered only one test, and 52% ordered up to five tests. However, a small proportion ordered > 50–100 tests/year (Figure 3A). Thus, the top three requestors ordered 7% of all tests, and 98 requestors ordered > 500 tests, that is, 1.2% of requestors ordered 44% of the total tests.

Figure 3.

Panel A: distribution of tests requested by individual requestors. The label above each bar shows the number and percentage of requestors (e.g., 1969 ordered one test, and 2022 ordered 2–5 tests). Panel B: total number of test results by individual requestors classified by 25OHD level (e.g., three ordered > 5000 tests of which 14% were ≤ 25 nmol/L (total 21,092 tests). Panel C: number of people having at least one repeated test from the same requestor (e.g., 23141 had one repeated test (i.e., two tests total) from the same requestor, 6237 had two repeated tests (three total) ordered by the same requestor). Panel D: total number of tests in individuals limited to repeated tests from the same requestor by number of repeated tests and classified by 25OHD level (e.g., 6237 individuals had two repeated tests ordered by the same requestor for a total of 12474 tests of which 2.7% were ≤ 25 nmol/L; 24 individuals had 20+ repeated tests by the same requestor for a total of 701 tests [range 21–50] of which 1.6% were ≤ 25 nmol/L).

Over the 20 years, 41% of requestors ordered < 1 test/year on average, while 899 (12%) requested > 1 test/month. The frequent requestors were not more likely to order tests with 25OHD ≤ 25 or ≤ 50 nmol/L (Figure 3B). Figure 3C,D shows the same analyses but restricted to individuals with repeated tests by the same requestor. The majority of individuals (67%) with multiple tests ordered by the same requestor had only 1 repeat test for a total of two tests in the 20‐year period, but 24 (0.1%) had > 20 tests. Figure 3D shows that the majority of the repeated tests by the same requestor in an individual are not low. Only 2.8% of repeated tests by the same requestor for an individual were ≤ 25 nmol/L, and likewise only 1.3% of tests in people with at least 10 repeated tests by the same requestor were ≤ 25 nmol/L.

Information on speciality was available for 88% of requests that originated from 49% of requestors. There were 73 different specialist categories: 37 adult, 34 paediatric, primary care, and refugee screening. Thus, 275,118 tests were requested by 3749 individual requestors with available specialty data for these categories. Figure 4A shows the number of requests by anonymised individual requestors identified by their speciality by 25OHD categorisation, restricted to those with > 1500 requests in the dataset. The largest requestor was a service that performs medical checks on refugees newly arrived in New Zealand. Of the remaining top 10 requestors, six were primary care doctors and three bariatric surgeons. Figure 4B shows the number of repeated tests in individuals by the same requestor with at least 330 repeated test requests.

Figure 4.

Panel A: the number of 25OHD tests for the largest volume requestors, identified by their speciality, and by the 25OHD category. For example, the second largest volume requestor was in primary care and requested 5328 tests. Panel B: the same analysis restricted to repeated tests in the same individual by the same requestor. For example, one bariatric surgeon requested 3439 25OHD tests in individuals with at least two tests requested by that bariatric surgeon. Panel C: the number of 25OHD tests and Panel D repeated tests in the same individual by the same requestor by the largest volume requestors grouped into their speciality and by 25OHD category. For example, nine bariatric surgeons requested 19721 tests (C), and 10,260 of these were by the same requestor in the same individual (D). The number on each bar in (A) and (B) is the number of tests, and in (C) and (D) is the number of requestors per specialty. Note in panel C and D, the specialities with the highest number of requests are plotted separately and on the right y‐axis.

The largest volume requestors did not have higher rates of low 25OHD results. Requestors with more than 500 requests had a median of 8.7% of 25OHD ≤ 25nmol/L (range 0.3%–34% compared with a median of 15.4% (range 0.3%–100%) for the entire cohort of requestors. For repeated requests in an individual by the same requestor, high volume requestors with more than 100 requests had a median of 4.2% of 25OHD ≤ 25nmol/L (range (0.5%–23%) compared with a median of 11.1% (0.04%–100%) for the entire cohort of requestors who made repeated requests in an individual.

Figure 4C,D show the number of tests and repeated tests by speciality. Primary care doctors were the most common speciality and requested almost 50% of tests. The next most common requestors were bariatric surgeons, even though there were only nine requestors, and the Refugee Screening Service. For repeated tests in the same individual, primary care doctors and bariatric surgeons were again the most common requestors by far. Several paediatric subspecialties featured among the top requestors for all and repeated tests. We further grouped the specialties into four categories—primary care, refugee services, adult services and paediatric services ‐ and repeated the analyses. Supporting Information S1: Figure 1 shows that paediatrics requested a disproportionately high number of overall and repeated tests compared to the size of the population they serve (in 2023, the proportion of the New Zealand population < 15 years was 18.5%, www.stats.govt.nz).

3.3. Modelling Effects of Further Restrictions

In 2012 restrictions for 25OHD testing were introduced for adult patients in Auckland. However, Table 1 shows that targeting of testing was still suboptimal: most tests results were at least 50 nmol/L, and only about 10% were ≤ 25 nmol/L. We explored what effects simple restrictions might currently have on subsequent testing. Table 2 shows the effects of various strategies based around time since previous test and the previous 25OHD result. Simply restricting testing to people without a 25OHD measurement in the last year would avoid 58% of repeat tests and 20% of all tests, and only 4% of avoided tests would be ≤ 25 nmol/L. Likewise, restricting repeat testing if the previous result was > 50 nmol/L would avoid 53% of repeated tests with only 2.2% of avoided tests being ≤ 25 nmol/L. Applying the combination of time since previous test (1 y) and previous result > 50 nmol/L increased the number of avoided tests to 81% while the proportion of avoided tests ≤ 25 nmol/L remained at a very low level (3.8%). If the criteria were made more stringent by increasing the time interval restriction and decreasing the 25OHD threshold, the proportion of tests avoided increased until nearly all repeat tests were avoided while the proportion of avoided tests with 25OHD ≤ 25 nmol/L remained low (3.5%–5.4%).

Table 2.

Potential repeated 25OHD tests avoided with different restriction strategies.

Strategy‐ decline request if:	Tests avoided (n)	Proportion of repeated tests (%)	Proportion of total tests (%)	Avoided tests with 25OHD ≤ 25 nmol/L n (%)
Test within 1 y	71,454	58	20	2741 (3.8)
Test within 2 y	94,080	77	26	4004 (4.3)
Test within 5 y	111,440	91	31	5595 (5.0)
Previous 25OHD:
> 100 nmol/L	10,766	9	3	87 (0.8)
> 75 nmol/L	29,503	24	8	339 (1.1)
> 50 nmol/L	64,472	53	18	1419 (2.2)
> 25 nmol/L	104,770	85	29	4212 (4.0)
Test 1 y & 25OHD > 75 nmol/L	84,472	69	23	2990 (3.5)
Test 1 y & 25OHD > 25 nmol/L	115,883	95	32	5525 (4.8)
Test 2 y & 25OHD > 75 nmol/L	101,232	83	28	4198 (4.1)
Test 2 y & 25OHD > 25 nmol/L	118,669	97	33	5952 (5.0)
Test 5 y & 25OHD > 75 nmol/L	113,889	93	31	5686 (5.0)
Test 5 y & 25OHD > 25 nmol/L	120,863	99	33	6481 (5.4)

Note: Tests avoided means the total number of tests that would have been avoided if the specific criteria had been used‐ for example, 71,454 tests were done within 12 months of the previous test, and thus would have been 'avoided' under that strategy. Proportions refer to the proportion of repeated and total 25OHD measurements respectively.

4. Discussion

In these datasets totalling over 360,000 25OHD results, repeated testing within 20 y occurred in about one in four people, and the repeat tests accounted for about 34% of the total numbers of tests. A disproportionately high number of tests were done in a small number of individuals, and most repeat tests (77%) were done within 2 y of the previous tests. Only a very small proportion of repeated tests were clearly low (≤ 25 nmol/L), and conversely, when the first 25OHD was > 100 nmol/L, 73% of repeated tests were > 75 nmol/L and 40% were > 100 nmol/L. The majority of 25OHD tests were ordered by a small number of practitioners. However, the results of the tests they requested were not more likely to be low than those requested by practitioners who ordered few tests. A small number of people had annual or 2‐yearly requests by the same requestor, even though test results in this group of people were very unlikely to be low. The majority of requests and repeated requests came from a small number of specialties: primary care, bariatric surgery, several paediatric specialties, and a few adult specialties.

Previously, we reported that the majority of 25OHD measurements do not identify people with low 25OHD who are at risk of complications of vitamin D deficiency [7]. In 2012 restrictions for 25OHD testing were introduced in Auckland because of a 380% increase in test requests between 2000 and 2010, with a resulting cost increase to > $1 million dollars/year [5, 6]. The introduction of restrictions aimed at better targeting of individuals who are likely to be vitamin D deficient have only been partially and temporarily successful. Overall, the number of requests decreased substantially after the restrictions were introduced, but the proportions of individuals with 25OHD ≤ 25 nmol/L remains at about 10%, and about 2/3 have 25OHD > 50 nmol/L, suggesting targeting of individuals at risk of vitamin D deficiency remains very poor. Total numbers of tests remained stable for about 5 years after restrictions were introduced but then rose exponentially, mirroring the rise in testing that led to the restrictions being placed initially.

The current analyses extend these findings to repeated 25OHD measurements and offer some potential approaches to improving testing. Most repeated tests are done in a relatively small group of people within 1 or 2 years of the first test. People selected for repeated testing do not have markedly lower 25OHD levels on their first test compared with people who do not have repeated tests, and the repeated tests themselves tend to be higher than the baseline levels. In New Zealand, it is recommended that people at risk of vitamin D deficiency do not need a 25OHD measurement before supplements are started because it is not cost‐effective [9, 10]. Likewise, people taking vitamin D supplements do not need a measurement of 25OHD, because it is unlikely that the level would be low or that the result would change management [9, 10]. A recent joint Endocrine Society Guideline statement has also advocated against routine measurement of 25OHD, measurement in individuals at high risk of vitamin D deficiency (vs. empiric supplementation), or in measurement in individuals taking vitamin D supplements [11]. It is therefore hard to understand why so many baseline tests are > 25 nmol/L, why repeat tests are requested, why they are requested so frequently, and how all these test would change patient management. However, it is important to acknowledge that the new Endocrine Society Guidelines were not released until 2024, and it will take time for this guidance to be reflected in clinical practice.

Request volumes differed greatly among practitioners and their specialties. A relatively small number of requestors from a small range of specialties requested the majority of tests and requested them frequently and repeatedly in their patients. However, these requestors were not more likely to identify people with low 25OHD, suggesting that they were just requesting tests routinely rather than in a targeted fashion. Previously, we reported similar findings [5, 6]. A few requestors accounted for most of the test requests more than a decade ago [5, 6], but educational efforts to reduce testing were not successful, prompting the introduction of restrictions.

We suggest that further restrictions are likely the best option to reduce unnecessary or inappropriate testing. Previous restrictions for adults were only modestly effective and should be extended to children. With a fairly simple approach to restrictions based around the time since the previous test and the test result, it is possible to avoid > 80% of repeated tests, which would reduce the overall test burden by > 25%. However, restrictions based solely around recent previous test results are limited in their ability to improve testing effectiveness because repeat tests only accounted for 34% of the total tests in this dataset. For initial testing, we suggest that much more focus should be placed on identifying individuals with a clinical complication of vitamin D deficiency. For example, previously we have shown that most adults and children in Auckland, even with 25OHD ≤ 25 nmol/L, do not have an associated biochemical abnormality [12, 13]. This suggests that aiming to identify individuals with 25OHD less than the threshold of 25 nmol/L, which is widely considered as indicating vitamin D deficiency, will identify few people with clinical complications of vitamin D deficiency because the rates are very low, even in groups considered high‐risk. Similar findings were reported from a study in Aberdeen, Scotland [14].

A simple approach focused on the identification of patients at risk of clinically relevant vitamin D deficiency would be to restrict testing to individuals with at least two out of three of low serum calcium, high ALP, or high PTH. Serum calcium and ALP are inexpensive, readily available tests that it seems reasonable to check first‐line in patients with symptoms suggestive of osteomalacia or severe vitamin D deficiency. Abnormal test results would identify patients at risk of clinically relevant vitamin D deficiency requiring investigation and treatment. Normal test results would minimise the need for 25OHD testing and the whole approach would encourage appropriate empiric vitamin D supplementation in at risk groups [11]. Table 1 shows the proportions of tests with associated biochemical testing and abnormal results in these data. Of the 15,622 25OHD tests with associated measurements of serum calcium, ALP and PTH, 80 (0.5%) had three abnormal results, and 834 (5.3%) had two abnormal results. Of the 74,383 25OHD tests with associated measurements of two of serum calcium, ALP, and PTH, 878 (1.2%) had at least two abnormal results, but 94% had two normal results meaning they would not meet the criteria for 25OHD testing. Addition of simple biochemical screening before 25OHD measurement is a simple step that is likely to substantially reduce 25OHD requests.

Elsewhere implementation of guidance and restrictions on testing have usually been successful but the benefits short‐lived. Introduction of testing guidance for general practitioners without formal restrictions produced a small, approximately 15%, reduction in testing in Australia [8], and a similar effect in Copenhagen, Denmark [15]. Implementation of a pop‐up box at time of testing reduced tests by 25% in the Capital Region of Denmark [16], and 27% in Ontario [17]. However, state‐wide restrictions of public funding of tests in Ontario had a much greater effect (82% reduction) [4], mirroring our own experience in Auckland of > 80% reduction in testing. However, in all these examples the reductions have been relatively short‐lived with testing returning to baseline levels within a few years in Australia [8] or 10–15 years in Ontario [17] and Auckland (current data). The most likely reason for the failure of restrictions to produce long‐term change is that the requesting clinicians disagree with restrictions and request the test even though the patient does not qualify. Requesting clinicians do not have to prove that their patient meets the criteria, they simply have to state that they do. Implementing systematic auditing to confirm eligibility‐ rather than relying on requestor declaration—would likely be time‐consuming and impractical given the volume of requests. Some requestors might feel that 25OHD measurement is required to identify individuals that will not benefit from supplementation because large‐scale trials indicate that vitamin D supplementation in vitamin D sufficient cohorts does not prevent major cardiovascular events or reduce overall cancer incidence [18]. Secondly, vitamin D is a subject of intense media focus, and patient expectation, for example, in the Covid 19 epidemic, might have led to pressure for testing [19]. Thirdly, clinicians might be unaware or dismissive of the costs of testing, or may be unaware that their practice of large volume vitamin D requesting places them as an outlier compared to their peers. Regardless of the exact reasons, it seems likely that tighter and specific restrictions and stricter enforcement of them are probably necessary to produce sustained changes.

We propose that the aim of a strategy for vitamin D testing should be to identify patients at highest risk of the complications of vitamin D deficiency. Thus, for a person without previous testing, 25OHD measurements could be restricted to (1) patients not taking vitamin D supplements, (2) patients with clinical or radiologic features of rickets or osteomalacia, (3) patients suspected to have vitamin D deficiency with at least two out of three of abnormal tests for serum calcium, ALP, or PTH, (4) patients undergoing investigations in specialist care for disorders of calcium and phosphate metabolism. For repeated 25OHD measurements, the above restrictions should apply plus (1) the patient has not had a 25OHD measurement within the past 2 y and (2) the previous 25OHD was not > 50 nmol/L. Patients otherwise at high risk who do not meet these criteria could be treated empirically and not tested [11]. There will be individual cases that do not meet these criteria for whom testing is clearly indicated, and these cases should be discussed with a chemical pathologist or endocrinologist directly.

One unique situation is the refugee medical centre which performs medical checks on refugees newly arrived in New Zealand. That service requested 9903 initial tests and 715 repeated tests, of which 14% and 30% respectively were ≤ 25 nmol/L, and 47% and 72% respectively were ≤ 50 nmol/L. A pragmatic approach for clinics seeing such population groups is not to test 25OHD but to routinely prescribe vitamin D supplements, give advice on sunshine exposure and the risk of vitamin D deficiency, and arrange suitable follow‐up in several months where the decision can be made to continue or discontinue the supplements.

It is important to recognise that the numbers of repeat tests in these analyses will be an underestimate, as some people we identified as having a single test only will have had a test prior to, or subsequent to, the period the datasets cover. Thus, our analyses are likely to be an underestimate of the numbers of people with repeat testing and the potential gains from further restrictions. Similarly, we did not have access to information on vitamin D supplementation which will influence individual 25OHD results. However, testing in patients taking vitamin D supplementation is contrary to New Zealand guidelines [9, 10].

In summary, vitamin D testing in the Auckland region remains predominantly in individuals at low risk of vitamin D deficiency, and requests are dominated by a relatively small number of individuals and specialities. Repeat tests are common and again do not clearly identify a subset of individuals with low 25OHD. The clinical value of repeat testing, especially within a relatively short interval (< 2 years) if the previous result was ≥ 50 nmol/L is very low. Previous attempts at educating high‐volume requestors were unsuccessful, and previous restrictions on 25OHD testing have only been partially successful and not sustained, suggesting that further restrictions should be put in place to reduce unnecessary tests, rationalise costs, and improve the targeting of people with low 25OHD at risk of complications of vitamin D deficiency.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.