https://orcid.org/0000-0002-5561-948X
ABSTRACT
Even in its early stages, chronic kidney disease (CKD) is associated with adverse pregnancy outcomes. The current guidelines for pregnancy management suggest identifying risk factors for adverse outcomes but do not mention kidney diseases. Since CKD is often asymptomatic, pregnancy offers a valuable opportunity for diagnosis. The present analysis attempts to quantify the cost of adding serum creatinine to prenatal screening and monitoring tests.
The decision tree we built takes several screening scenarios (before, during and after pregnancy) into consideration, following the hypothesis that while 1:750 pregnant women are affected by stage 4–5 CKD and 1:375 by stage 3B, only 50% of CKD cases are known. Prevalence of abortions/miscarriages was calculated at 30%; compliance with tests was hypothesized at 50% pre- and post-pregnancy and 90% during pregnancy (30% for miscarriages); the cost of serum creatinine (production cost) was set at 0.20 euros. A downloadable calculator, which makes it possible to adapt these figures to other settings, is available.
The cost per detected CKD case ranged from 111 euros (one test during pregnancy, diagnostic yield 64.8%) to 281.90 euros (one test per trimester, plus one post-pregnancy or miscarriage, diagnostic yield 87.7%). The best policy is identified as one test pre-, one during and one post-pregnancy (191.80 euros, diagnostic yield 89.4%).
This study suggests the feasibility of early CKD diagnosis in pregnancy by adding serum creatinine to routinely performed prenatal tests and offers cost estimates for further discussion.
What is already known about this subject?
Chronic kidney disease (CKD) is a risk factor for preeclampsia and up to 20% of women who have experienced preeclampsia are thought to have underlying CKD, which frequently fails to be diagnosed.
The European guidelines for routine prenatal care do not include serum creatinine testing to screen for CKD.
Universal screening for CKD in pregnancy is increasingly being advocated by the nephrology community.
What this study adds?
A cost-effectiveness analysis of screening for CKD in pregnancy by measuring serum creatinine.
What impact this may have on practice or policy?
Early CKD diagnosis could improve mother and fetal outcomes during pregnancy.
Early CKD diagnosis could improve women's health in the long term.
INTRODUCTION
Chronic kidney disease (CKD) is a risk factor for adverse pregnancy outcomes even in the early stages of the disease [1–3]. However, while the current European and US guidelines for the management of low-risk pregnancy include a search for major risk factors such as diabetes and hypertension, they do not include measuring kidney function [4–6]. While most guidelines recommend including questions on diabetes, hypertension and thyroid diseases, none specifically mentions determining whether a patient has a history of kidney disease or suggest testing kidney function before or at the beginning of pregnancy, thus failing to acknowledge that CKD is often asymptomatic and may not be found unless searched for [7–9]. Although dipstick urinalysis, the only test routinely prescribed in pregnancy, serves to identify kidney diseases characterized by proteinuria, its limits as a sole marker of kidney disease are well known and this test is probably more important, during the second half of pregnancy, for identifying preeclampsia [4–7]. Attention to microscopic haematuria can increase the diagnostic yield of urinalysis, but, since it is found in up to 20% of uneventful pregnancies, its association with CKD is loose in this setting [10].
By definition, CKD encompasses any abnormality of kidney structure or function [glomerular filtration rate (GFR) measured or calculated] or blood and urine composition, including alteration in blood or electrolytes [11]. Thus, CKD diagnosis requires a combined assessment of kidney function and electrolytes, renal imagery and urinalysis. A urinalysis alone cannot detect reduced kidney function, kidney malformations and scars, interstitial nephropathies and glomerulonephritis in remission unless they are associated with proteinuria, which is not common in these conditions [12].
Many, if not most kidney diseases are asymptomatic and awareness of having CKD is reported to be low, reaching 80% in patients treated in nephrology, but often lower than 10% in the general population and reaching at most 50% in the most advanced CKD stages (stage 3–5) [13–17]. Pregnancy is often the first time that biochemical tests are performed on an asymptomatic young woman; in this context, assessment of the kidney function can identify patients with asymptomatic CKD and indicate pregnancies that require particular attention, allowing the timely provision of care and better outcomes [18–21].
On the occasion of the 2018 World Kidney Day dedicated to women and kidney diseases, the integration of serum creatinine, a simple, low-cost, widespread and pivotal marker of kidney function, in the work-up for pregnant patients was suggested as a potential tool for improving women's health [22]. To date, this suggestion has not been widely followed.
In the present study, we calculated, modelling data available from the literature, how much detecting new CKD cases with reduced kidney function would cost, adding serum creatinine to pregnancy screening and monitoring tests, in several scenarios. Due to the changes in glomerular filtration rate in pregnancy, creatinine levels must be interpreted with caution; acknowledging this, we conservatively modelled our decision tree focussing on the detection of cases with a relevant reduction in kidney function (CKD stage 3b onwards) [23, 24].
MATERIALS AND METHODS
Baseline hypotheses
The model discussed here is based on the following series of assumptions (Table 1):
The prevalence of CKD in pregnant women is the same as in women of childbearing age. We relied on Davison's classic estimate that sets CKD at about 3% in this population and estimates the prevalence of ‘severe’ CKD to be 1:750 [25]. This was considered as synonymous with a prevalence of CKD stages 4–5 (0.13333% of the population). As for stage 3, considering that the prevalence is usually estimated to be 4–10 times higher than CKD stages 4–5, we hypothesized that it was evenly divided between stages 3a and 3b and focussed only on stage 3b [estimated GFR (eGFR) 44–30 mL/min], to account for the difficulty in detecting lesser degrees of a creatinine increase in pregnancy [23, 24]. We conservatively estimated prevalence as double that of stages 4–5 (i.e. 0.26667% of the population). The estimate is conservative and in line with a large recent meta-analysis [26].
Targeting the model to the detection of cases with reduced kidney function, we hypothesized that a patient with CKD stage 3b before pregnancy (eGFR 30–44 mL/min) would remain in stage 3, possibly shifting to 3a (eGFR 59–45 mL/min), during pregnancy. This choice is quite conservative: we did not consider the possibility of detecting earlier stages (1–3a), if serum creatinine were to be adjusted for pregnancy and urinalysis and hypertension were considered [24, 25].
Having set the detection threshold at relatively low eGFR levels, we considered that a further 10% of cases would escape detection because of laboratory errors (detection rate of 90%). Laboratory errors were broadly defined as any defect from ordering tests to reporting and interpreting results [36]. With respect to serum creatinine, a 10% variability was retained after discussion with the head of our laboratory, situated in one of the three largest non-university hospitals in France. This estimate is in keeping with a recent study on delta checks of serum creatinine in the laboratory context, highlighting the overall high reliability of serum creatinine laboratory assessment [37]. This is a very conservative, empiric estimate that includes pre-analytical errors (such as non-fasting test, hyper-hydration or under-hydration); laboratory variability (analyser-related); unacknowledged differences among different laboratory tests employed (Jaffé, enzymatic).
We did not consider ‘false positives’ because, even with a 20% decrease in GFR (corresponding to the highest variability of serum creatinine tests), these cases would have an eGFR < 60 mL/min and warrant evaluation, in the very conservative diagnostic setting we had chosen.
Our proposal regards the inclusion of serum creatinine in standard outpatient controls that are advised in the absence of signs and symptoms of the disease. In the case of severe infection, persistent fever, severe hyperemesis or other diseases potentially causing pregnancy-related acute kidney injury (AKI), serum creatinine should be routinely measured and followed up at least to full normalization. AKI and CKD are intrinsically linked and attention to AKI can lead to early CKD diagnosis [20–22]. As we lacked epidemiological data, we did not consider this event in the present model.
We did not consider the diagnostic yield of urinalysis in detecting forms of CKD characterized by proteinuria, nor did we consider haematuria, due to its high prevalence in pregnancy [10]. While we used conservative figures for all other items, given the lack of data on the prevalence of CKD stages 3b–5 and proteinuria in pregnancy, we did not try to adjust for the prevalence of cases that would have been detected by proteinuria.
We assumed that 50% of the patients would not be aware of the presence of CKD [13–16], a figure that is probably underestimated, as recent studies in the US set CKD unawareness at higher than 90% of cases, even in the late stages of the disease and found higher levels of awareness only in patients being followed in a nephrology setting [13–16]. No data exist on the awareness of CKD in younger patients, but it is known that awareness increases with kidney function impairment and, since we focussed on relatively late CKD stages, we once more chose a very conservative figure.
The abortion/miscarriage rate, adherence to tests and detection rate (based on standard laboratory error for serum creatinine) were modelled on the data in the literature (Table 1).
We chose a rounded cost of 0.20 euros per serum creatinine test, considering, in the European setting, that the test would be added to the ones fully covered by the patient's healthcare system and performed consensually (costs per test evaluated at 0.10–0.30 euros). The cost per diagnosis, therefore, represents the direct cost covered by the healthcare system. Since we hypothesized that the test would be performed consensually with ones that had already been prescribed during the course of pregnancy, indirect costs borne by the patient (loss of working time, transportation, etc.) or by the structure (personnel for blood sampling, etc.) were not added [38].
Main assumptions for the calculation of costs and references
| Item | Chosen values | Note |
|---|---|---|
| CKD prevalence in childbearing age: all cases all stages | 3% (1–5%) | Classic estimate, in line with recent meta-analyses [25, 26, 27]. May be higher in developing countries [18, 19, 21]. |
| CKD prevalence in childbearing age: stages 4–5 | 0.133% (0.05–0.50%) 1:750 | Classic estimate, in line with recent meta-analyses [25, 26]. May be higher in developing countries [18, 19, 21]. |
| CKD prevalence in childbearing age: stage 3b | 0.267% (0.1–1%) | Extrapolation from [25, 26]. Twice as frequent as stages 4–5. May be higher in developing countries. |
| Known CKD | 50% (5–80%) | Literature data. Lower in early CKD stages, reaches 80% only in nephrology settings, but is reported to be lower than 50% even in late stages, in patients not followed in nephrology [13–17]. Data on women of childbearing age are lacking. |
| Detection rate of CKD | 90% stages 3b–5 (80–95%) | Detection rate conservatively assessed, considering the variability in creatinine assessment [28]. Detection rate may be higher before and after pregnancy. Adjustment is needed for correct interpretation in pregnancy, but we did not consider the recent indications on adjustment, as they are not fully acknowledged in clinical practice [23, 24]. |
| Cost of serum creatinine tests | 0.10–0.20 euros per added test 1.00–1.50 euros per self-standing test | Cost of tests depends on technique (Jaffe, enzymatic) and number of tests per laboratory. Average data from hospital laboratories in the setting of study (unpublished data from the hospital management). |
| Number of tests per pregnancy | 1–5 tests | According to prescription: pre-; each trimester; post-pregnancy (empirically tested, according to the pregnancy control schedules [4–7]. |
| Adherence to tests | 50% pre- and post- pregnancy (20–60%); miscarriages: 30% (20–40%); at least one test in pregnancy (70–95%) | Literature data [29, 30–32]. May be lower in developing countries, or where the cost of the test is not reimbursed. |
| Abortion–miscarriage rate | 30% (10–45%) | Literature data [33–35]. Prevalence merges voluntary pregnancy terminations (20–50%) and miscarriages (8–15%). The first depends significantly on cultural and religious settings, and varies from country to country. |
| Item | Chosen values | Note |
|---|---|---|
| CKD prevalence in childbearing age: all cases all stages | 3% (1–5%) | Classic estimate, in line with recent meta-analyses [25, 26, 27]. May be higher in developing countries [18, 19, 21]. |
| CKD prevalence in childbearing age: stages 4–5 | 0.133% (0.05–0.50%) 1:750 | Classic estimate, in line with recent meta-analyses [25, 26]. May be higher in developing countries [18, 19, 21]. |
| CKD prevalence in childbearing age: stage 3b | 0.267% (0.1–1%) | Extrapolation from [25, 26]. Twice as frequent as stages 4–5. May be higher in developing countries. |
| Known CKD | 50% (5–80%) | Literature data. Lower in early CKD stages, reaches 80% only in nephrology settings, but is reported to be lower than 50% even in late stages, in patients not followed in nephrology [13–17]. Data on women of childbearing age are lacking. |
| Detection rate of CKD | 90% stages 3b–5 (80–95%) | Detection rate conservatively assessed, considering the variability in creatinine assessment [28]. Detection rate may be higher before and after pregnancy. Adjustment is needed for correct interpretation in pregnancy, but we did not consider the recent indications on adjustment, as they are not fully acknowledged in clinical practice [23, 24]. |
| Cost of serum creatinine tests | 0.10–0.20 euros per added test 1.00–1.50 euros per self-standing test | Cost of tests depends on technique (Jaffe, enzymatic) and number of tests per laboratory. Average data from hospital laboratories in the setting of study (unpublished data from the hospital management). |
| Number of tests per pregnancy | 1–5 tests | According to prescription: pre-; each trimester; post-pregnancy (empirically tested, according to the pregnancy control schedules [4–7]. |
| Adherence to tests | 50% pre- and post- pregnancy (20–60%); miscarriages: 30% (20–40%); at least one test in pregnancy (70–95%) | Literature data [29, 30–32]. May be lower in developing countries, or where the cost of the test is not reimbursed. |
| Abortion–miscarriage rate | 30% (10–45%) | Literature data [33–35]. Prevalence merges voluntary pregnancy terminations (20–50%) and miscarriages (8–15%). The first depends significantly on cultural and religious settings, and varies from country to country. |
Main assumptions for the calculation of costs and references
| Item | Chosen values | Note |
|---|---|---|
| CKD prevalence in childbearing age: all cases all stages | 3% (1–5%) | Classic estimate, in line with recent meta-analyses [25, 26, 27]. May be higher in developing countries [18, 19, 21]. |
| CKD prevalence in childbearing age: stages 4–5 | 0.133% (0.05–0.50%) 1:750 | Classic estimate, in line with recent meta-analyses [25, 26]. May be higher in developing countries [18, 19, 21]. |
| CKD prevalence in childbearing age: stage 3b | 0.267% (0.1–1%) | Extrapolation from [25, 26]. Twice as frequent as stages 4–5. May be higher in developing countries. |
| Known CKD | 50% (5–80%) | Literature data. Lower in early CKD stages, reaches 80% only in nephrology settings, but is reported to be lower than 50% even in late stages, in patients not followed in nephrology [13–17]. Data on women of childbearing age are lacking. |
| Detection rate of CKD | 90% stages 3b–5 (80–95%) | Detection rate conservatively assessed, considering the variability in creatinine assessment [28]. Detection rate may be higher before and after pregnancy. Adjustment is needed for correct interpretation in pregnancy, but we did not consider the recent indications on adjustment, as they are not fully acknowledged in clinical practice [23, 24]. |
| Cost of serum creatinine tests | 0.10–0.20 euros per added test 1.00–1.50 euros per self-standing test | Cost of tests depends on technique (Jaffe, enzymatic) and number of tests per laboratory. Average data from hospital laboratories in the setting of study (unpublished data from the hospital management). |
| Number of tests per pregnancy | 1–5 tests | According to prescription: pre-; each trimester; post-pregnancy (empirically tested, according to the pregnancy control schedules [4–7]. |
| Adherence to tests | 50% pre- and post- pregnancy (20–60%); miscarriages: 30% (20–40%); at least one test in pregnancy (70–95%) | Literature data [29, 30–32]. May be lower in developing countries, or where the cost of the test is not reimbursed. |
| Abortion–miscarriage rate | 30% (10–45%) | Literature data [33–35]. Prevalence merges voluntary pregnancy terminations (20–50%) and miscarriages (8–15%). The first depends significantly on cultural and religious settings, and varies from country to country. |
| Item | Chosen values | Note |
|---|---|---|
| CKD prevalence in childbearing age: all cases all stages | 3% (1–5%) | Classic estimate, in line with recent meta-analyses [25, 26, 27]. May be higher in developing countries [18, 19, 21]. |
| CKD prevalence in childbearing age: stages 4–5 | 0.133% (0.05–0.50%) 1:750 | Classic estimate, in line with recent meta-analyses [25, 26]. May be higher in developing countries [18, 19, 21]. |
| CKD prevalence in childbearing age: stage 3b | 0.267% (0.1–1%) | Extrapolation from [25, 26]. Twice as frequent as stages 4–5. May be higher in developing countries. |
| Known CKD | 50% (5–80%) | Literature data. Lower in early CKD stages, reaches 80% only in nephrology settings, but is reported to be lower than 50% even in late stages, in patients not followed in nephrology [13–17]. Data on women of childbearing age are lacking. |
| Detection rate of CKD | 90% stages 3b–5 (80–95%) | Detection rate conservatively assessed, considering the variability in creatinine assessment [28]. Detection rate may be higher before and after pregnancy. Adjustment is needed for correct interpretation in pregnancy, but we did not consider the recent indications on adjustment, as they are not fully acknowledged in clinical practice [23, 24]. |
| Cost of serum creatinine tests | 0.10–0.20 euros per added test 1.00–1.50 euros per self-standing test | Cost of tests depends on technique (Jaffe, enzymatic) and number of tests per laboratory. Average data from hospital laboratories in the setting of study (unpublished data from the hospital management). |
| Number of tests per pregnancy | 1–5 tests | According to prescription: pre-; each trimester; post-pregnancy (empirically tested, according to the pregnancy control schedules [4–7]. |
| Adherence to tests | 50% pre- and post- pregnancy (20–60%); miscarriages: 30% (20–40%); at least one test in pregnancy (70–95%) | Literature data [29, 30–32]. May be lower in developing countries, or where the cost of the test is not reimbursed. |
| Abortion–miscarriage rate | 30% (10–45%) | Literature data [33–35]. Prevalence merges voluntary pregnancy terminations (20–50%) and miscarriages (8–15%). The first depends significantly on cultural and religious settings, and varies from country to country. |
Building the model
Considering that the assumptions listed above, though justified by the scientific literature, are subject to variations based on socio-economic and cultural settings, in order to be consistent with the natural history of the disease and the standards of care in European countries and at the same be adaptable to other settings and contexts, we decided to use a simplified flexible model derived from a decision-tree approach [39]. The model was developed using Microsoft Excel 2016 and is downloadable as Supplementary data. The spreadsheet is modifiable and makes it possible to adapt costs and baseline hypotheses to different clinical and socio-economic models of care.
The robustness of the cost-effectiveness analysis was tested through a deterministic one-way sensitivity analysis model, limited to the least expensive/most efficient scenarios.
The model identifies the variation in costs per case detected according to variations in the following factors: CKD prevalence in childbearing age for stages 4–5; the prevalence of known CKD; detection rate of CKD; cost of serum creatinine tests; adherence to tests; abortion rate (see Table 1).
Results are presented as a tornado chart: for each variable considered, the chart shows the estimate of the base, lowest and highest cost, assuming that all the other variables are stable (i.e. they remain at baseline value) [40, 41].
RESULTS
Identification of the number of cases to be detected
According to the data in the literature (Table 1), we hypothesized there would be 30 000 cases with CKD per million pregnant women. Considering a prevalence of stages 3b–5 of 0.40% (4000 cases), half of which were known, we modelled the detection of 2000 cases with unknown CKD per million pregnancies, rounded ignoring correction for the sake of simplicity (1 000 000 – 2000 known cases = 998 000; 1996 cases to be detected). The assumptions leading to the calculation of the number of cases that would be found, reported in Table 1, are graphically plotted in Figure 1.
Flow-chart of the main assumptions leading to the identification of the number of cases that would be found.
Analysis of different scenarios
The assumptions reported in Table 1 and Figure 1 were applied to six different scenarios combining tests before, during and after pregnancy (Table 2).
Different scenarios and their diagnostic yield
| Scenario 1: | Scenario 2: | Scenario 3: | |
|---|---|---|---|
| N tests | One test performed during pregnancy (prescribed in the first trimester) | One test in the first trimester, repeated if normal, in each trimester | One test pre-pregnancy, 1 test during pregnancy |
| N cases detected | 1296 | 1552 | 1613 |
| Diagnostic yield (expected 2000) | 64.8% | 77.6% | 80.7% |
| Scenario 4: | Scenario 5: | Scenario 6: | |
| N tests | As scenario 1 plus 1 test post-pregnancy or post-miscarriage | As scenario 2 plus 1 test post-pregnancy or post-miscarriage | As scenario 3 plus 1 test post-pregnancy or post-miscarriage |
| N cases detected | 1613 | 1754 | 1788 |
| Diagnostic yield (expected 2000) | 80.7% | 87.7% | 89.4% |
| Scenario 1: | Scenario 2: | Scenario 3: | |
|---|---|---|---|
| N tests | One test performed during pregnancy (prescribed in the first trimester) | One test in the first trimester, repeated if normal, in each trimester | One test pre-pregnancy, 1 test during pregnancy |
| N cases detected | 1296 | 1552 | 1613 |
| Diagnostic yield (expected 2000) | 64.8% | 77.6% | 80.7% |
| Scenario 4: | Scenario 5: | Scenario 6: | |
| N tests | As scenario 1 plus 1 test post-pregnancy or post-miscarriage | As scenario 2 plus 1 test post-pregnancy or post-miscarriage | As scenario 3 plus 1 test post-pregnancy or post-miscarriage |
| N cases detected | 1613 | 1754 | 1788 |
| Diagnostic yield (expected 2000) | 80.7% | 87.7% | 89.4% |
Different scenarios and their diagnostic yield
| Scenario 1: | Scenario 2: | Scenario 3: | |
|---|---|---|---|
| N tests | One test performed during pregnancy (prescribed in the first trimester) | One test in the first trimester, repeated if normal, in each trimester | One test pre-pregnancy, 1 test during pregnancy |
| N cases detected | 1296 | 1552 | 1613 |
| Diagnostic yield (expected 2000) | 64.8% | 77.6% | 80.7% |
| Scenario 4: | Scenario 5: | Scenario 6: | |
| N tests | As scenario 1 plus 1 test post-pregnancy or post-miscarriage | As scenario 2 plus 1 test post-pregnancy or post-miscarriage | As scenario 3 plus 1 test post-pregnancy or post-miscarriage |
| N cases detected | 1613 | 1754 | 1788 |
| Diagnostic yield (expected 2000) | 80.7% | 87.7% | 89.4% |
| Scenario 1: | Scenario 2: | Scenario 3: | |
|---|---|---|---|
| N tests | One test performed during pregnancy (prescribed in the first trimester) | One test in the first trimester, repeated if normal, in each trimester | One test pre-pregnancy, 1 test during pregnancy |
| N cases detected | 1296 | 1552 | 1613 |
| Diagnostic yield (expected 2000) | 64.8% | 77.6% | 80.7% |
| Scenario 4: | Scenario 5: | Scenario 6: | |
| N tests | As scenario 1 plus 1 test post-pregnancy or post-miscarriage | As scenario 2 plus 1 test post-pregnancy or post-miscarriage | As scenario 3 plus 1 test post-pregnancy or post-miscarriage |
| N cases detected | 1613 | 1754 | 1788 |
| Diagnostic yield (expected 2000) | 80.7% | 87.7% | 89.4% |
The cost per new diagnosis, calculated according to our basic assumptions (Table 1, Figure 1), considering a production cost of 0.20 euros per creatinine test, is reported in Figure 2, which is also the graphic output of the calculation sheet available online (supplementary data). While, understandably, the simplest scenario (one test during pregnancy) is the least expensive, it detects less than two-thirds of the cases; the best performance combines one test before pregnancy, one test during pregnancy and one test after delivery (expected, based on our assumptions, to detect almost 90% of the cases).
Cost per new diagnosis, assuming a production cost of serum creatinine of 0.20 euros per test.
The flow chart of scenario 6, which has the highest diagnostic yield for an intermediate cost (191 euros per new diagnosis), is reported in Figure 3.
Flow chart displaying the diagnostic yield at each step in the most favourable scenario 6, combining tests before, during and after pregnancy.
Sensitivity analysis
As reported in Figure 4, the data on which our model was built show a high degree of variability, which results from the high heterogeneity of the source studies. In terms of economic yield, the impact on cost is relevant and is mainly modulated, beyond the cost of the test itself, by the prevalence of unacknowledged CKD and by the prevalence of miscarriages and pregnancy terminations. In the input page of the downloadable calculator, shown in Figure 5, all assumptions are modifiable and changing them leads to variations in the output page.
Sensitivity analysis for scenario 6. CKD, chronic kidney disease. The input parameters are reported in Table 1.
Input page of the downloadable calculator.
DISCUSSION
Early diagnosis of CKD is an often mentioned and rarely pursued goal. No single test is able to detect all cases of CKD, whose diagnosis often requires integration between blood and urine tests and renal imaging [12]. Therefore, a screening policy is deemed to represent a compromise between an interest in discovering as many cases as possible and the feasibility and cost of blood, urine and imaging tests.
Pregnancy is in a delicate balance between being a physiological phase of life that should be managed with the minimum of interference and a valuable opportunity to invest in the future health of mother and child. The evident advantages of a non-medicalized pregnancy are counterbalanced by the risk of missing the opportunity to diagnose non-symptomatic diseases, which can affect both pregnancy and the long-term health of the mother and offspring, of which CKD is probably the most frequent and clinically relevant one. Pregnancy complications are closely associated with the future development of chronic cardiovascular and kidney diseases [42–46], while CKD is associated with a higher risk of pregnancy complications [1–3] and preeclampsia has recently been associated with a high prevalence of CKD, as much as about 20%, found in post-natal assessment [29, 47–49]. While postnatal screening for kidney diseases is increasingly being suggested at least for patients who experience pregnancy complications [22, 29, 46–50], some data suggest that baseline kidney function, even in the normal range, may be associated with pregnancy outcomes, as pregnancy represents a sort of ‘stress test’ that can predict the development of future maternal diseases [51–54].
Adding serum creatinine to the pregnancy tests would make it possible to detect kidney diseases characterized by a reduction in kidney function that are not accompanied by proteinuria and would increase our understanding of the relationship between kidney function and pregnancy outcomes. The test is standardized and relatively inexpensive, is widely available and can easily be added to the set of laboratory analyses that are routinely prescribed [28]. The expenditure of adding serum creatinine to the tests already performed during pregnancy (albeit with subtle differences) in most European countries would result in an increase in expenditure by the healthcare system of 0.20 euros (one test only) to 1 euro per pregnancy (5 tests), in the settings in which the basic controls in pregnancy are free of charge (production costs: Table 1).
Due to the high heterogeneity of the assumptions derived from the literature, regarding both clinical elements that may be modulated by genetics and lifestyle, such as the prevalence of CKD and incidence of miscarriages and social ones, including adherence to prescribed controls and voluntary pregnancy terminations, or the prevalence of CKD unawareness, we decided to build a very simple flexible model, based on a decision tree, that would be adaptable to other settings (Figures1 and 5).
With the current assumptions, the cost per detected case of CKD stages 3b–5 ranges from about 100 to almost 300 euros and the best performance (about 90% diagnostic yield) is intermediate (about 200 euros per detected case) and combines one test before, one during and one after pregnancy (Table 2, Figure 3). The modular calculator allows changing the parameters, adapting them to different contexts. For instance, reducing the adherence to one test in pregnancy from 30 to 20% in miscarriages, and from 50 to 20% after pregnancy, would result, in scenario 1, in a decrease in diagnostic yield from 64.8 to 62.1% and from 80.7 to 79.2% in scenario 3; likewise, reducing the post-pregnancy tests from 50 to 25% would decrease the performance of scenario 6 from 89.4 to 85%.
This analysis, which has the strength of novelty, may help us determine the feasibility of a policy of universal screening in pregnancy, increasingly advocated at least by the nephrology community, but has several limits that should be addressed in future studies.
First, many of the assumptions derived from the literature are based on imprecise estimates and indirect evaluations. Secondly, we considered that the detected cases would not have been detected solely by urinalysis, acknowledging the importance of non-proteinuric CKD. The actual prevalence of such cases is unknown and only the systematic introduction of kidney function tests in pregnancy will make it possible to compare the diagnostic yield of urinalysis and serum creatinine assessments. This is also why we choose a conservative figure of 50% of known CKD; this figure can be modulated in the calculator, thus varying the cost per case detected (online calculator).
Thirdly, we considered that compliance would be stable during pregnancy, while it may vary across trimesters, and we considered that all pregnancy losses would occur in the first trimester. Furthermore, we did not consider over-diagnosis, due to laboratory variability and the inclusion of cases with pregnancy-related AKI. However, the model is based on the assumption that serum creatinine is dosed on the occasion of the usual tests performed in pregnancy and these are usually avoided in the presence of diseases potentially causing AKI, when extra tests are usually prescribed, including kidney function assessment. Furthermore, the 90% detection rate, employed in our model, is based upon empiric reasoning and is not validated in different contexts.
The advantages of early diagnosis of kidney disease are intuitive, at least for nephrologists, and a cost of about 200 euros for identifying a relatively advanced (stage 3b–5) and probably evolutive kidney disease seems reasonable, taking into account the extremely high costs of advanced CKD management or dialysis, estimated at about 75 000 euros per patient per year in the study setting, a figure that is not substantially different from what is recorded in most western countries [55, 56]. An indirect advantage of this approach is that it may contribute to establish normal creatinine values for pregnancy in different trimesters and to better understanding of the trajectories of CKD in pregnancy, by performing systematic measurements in case of kidney function reduction at the screening test. Standardization of serum creatinine measurement will, of course, be necessary.
We acknowledge, however, that we do not know the long-term impact of early CKD diagnoses. We know that not all kidney diseases are alike and that many of them do not progress over long periods of time [57, 58]. Furthermore, even if the indications of several nephrology societies, including ours, are in favour of early identification of all cases of CKD, no randomized study has assessed the advantages of such a demanding policy [59–61]. We do, however, know the effects of a late CKD diagnosis, which involves lost opportunities for individuals and for society [62–64].
This study suggests the feasibility of early CKD diagnosis in pregnancy and offers cost estimates for further discussion. Only prospective studies that systematically test serum creatinine in pregnancy can actually clarify the cost-effectiveness of the inclusion of this test in the pregnancy work-up and in improving the short- and long-term health of the mother and her offspring.
ACKNOWLEDGEMENTS
We thank Susan Finnel for her careful language editing.
AUTHORS’ CONTRIBUTIONS
Research idea and study design: G.B.P.; data acquisition: A.C., M.C., R.A., B.M., G.C., E.V.; data analysis/interpretation: A.C., M.T., G.B.P.; statistical analysis: A.C.; drafting: G.B.P.; final version: all authors. Each author contributed important intellectual content during manuscript drafting or revision, accepts personal accountability for their own contributions and agrees to ensure that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved.
FUNDING
No funding was received for this study; the Centre Hospitalier Le Mans supported editing and publishing expenses.
DATA AVAILABILITY STATEMENT
No new data were generated or analysed in support of this research.
CONFLICT OF INTEREST STATEMENT
There is no conflict of interest for any of the authors. The results presented in this paper have not been published previously in whole or in part.
APPENDIX
The group members of Kidney and Pregnancy Study Group of the Italian Society of Nephrology are Domenico Santoro, Gabriella Moroni, Linda Gammaro, Giuseppe Gernone and Bianca Covella.
Comments