Pregnancy has a favorable effect on the course of rheumatoid arthritis (RA), although the magnitude of this effect is equivocal because RA assessment tools have never been validated in pregnancy. The goal of this study was to assess how pregnancy influences the scoring of the Disease Activity Score in 28 joints (DAS28) and the Health Assessment Questionnaire (HAQ), and how both scores perform in pregnant patients with RA.
Thirty-two healthy women and 30 pregnant patients with RA were prospectively studied during pregnancy and at postpartum. At each trimester and postpartum the components of the DAS28 (global health [GH], erythrocyte sedimentation rate [ESR], and C-reactive protein level [CRP]) and HAQ scores were determined. Maximal influences of healthy pregnancy on each component of the DAS28 were calculated. The performances of different DAS28 scores and the HAQ were also determined in RA patients. Furthermore, variants of the HAQ were developed within the HAQ scoring rules.
The components of the DAS28 were influenced by healthy pregnancy, with average increases in DAS28 score of 0.22 (GH), 1.1 (ESR), and 0.25 (CRP). The DAS28 calculated with CRP (DAS28-CRP) and without GH performed the best in pregnant RA patients. In healthy pregnancy, the median HAQ increased to 0.50 in the third trimester and was reduced by the HAQ variants to 0.25.
Pregnancy considerably influences the scoring of the DAS28 and HAQ. RA disease activity in pregnant patients should preferably be calculated with DAS28-CRP without GH. Even with HAQ variants, influences of pregnancy on the assessment of functionality cannot be precluded.
Pregnancy has an important impact on rheumatoid arthritis (RA). It is the only natural situation in which affected women can experience complete remission of RA. During pregnancy, patients may experience signs and symptoms of amelioration of RA, followed by deterioration postpartum (1–3). The immunologic phenomena contributing to this favorable effect of pregnancy are of great interest in understanding the pathogenesis of RA.
Because remission has always been defined differently, the retrospective and prospective studies to date have shown substantial variations in disease activity during pregnancy (4–9). Criteria for remission were based variously on changes in clinical assessments, medication, and Health Assessment Questionnaire (HAQ) scores. These studies also used methods that were never validated during pregnancy.
To properly study the course of RA, disease activity and functioning of pregnant patients with RA should be studied prospectively throughout pregnancy using currently accepted scores for the assessment of RA, such as the modified Disease Activity Score including 28-joint counts (DAS28) (10, 11) and the HAQ (12, 13). However, these widely used RA assessment tools may not be valid for assessing RA during pregnancy. This is because pregnancy itself may influence not only the parameters of the DAS28, i.e., the visual analog scale (VAS) of global health (GH), erythrocyte sedimentation rate (ESR) (14), or C-reactive protein (CRP) level (15), but also the rating of some categories of the HAQ. For example, a 30-week pregnancy will most certainly cause some difficulties in functioning (16). Some adaptation or correction for the influence of pregnancy on currently accepted scores might therefore be warranted. The goal of our study was to investigate 1) how pregnancy itself influences the scoring of the DAS28 and HAQ and reliability of the DAS28 and HAQ as RA assessment tools in pregnancy, 2) how the HAQ can be adapted for use in pregnant patients with RA, and 3) how different DAS28 formulas and adaptations of HAQ perform during pregnancy in women with RA.
SUBJECTS AND METHODS
The Pregnancy-induced Amelioration of Rheumatoid Arthritis (PARA) study is a prospective, nationwide cohort study designed to investigate the amelioration of RA during pregnancy and the postpartum flare. The PARA study was started in 2002. In this study, healthy women serve as a reference group to determine the pregnancy-induced deviations of current scoring methods and laboratory tests. All participants gave their informed consent.
At the time of this study, the data collected on 30 consecutive full pregnancies of 30 patients with RA were available. All patients fulfilled the American College of Rheumatology (formerly the American Rheumatism Association) 1987 classification criteria for RA (17). Patients were visited at each trimester (9–12 weeks of gestation, 20 weeks, and 30 weeks) and twice postpartum (6 weeks and 12 weeks). At each visit patients filled out the HAQ and a VAS GH, and a standardized 28-joint count was performed. At the same time, ESR and CRP level were measured in blood samples.
Thirty-two healthy pregnant women were recruited from a midwifery practice between July 2002 and January 2004. Each woman was <13 weeks pregnant. The healthy women were visited at the same time points as the patients. The same assessments and laboratory tests were performed, except for the joint count.
The disease activity of patients can be defined by several DAS formulas. In general, a DAS is a statistically derived index combining tender joints, swollen joints, a laboratory parameter for inflammation (ESR or CRP), and global disease activity. In this study, the DAS28 was chosen as an assessment tool for disease activity.
Currently, 4 formulas are available to calculate the DAS28 (see Appendix A). In general, these 4 variants are composed of a 28-joint count for swelling and for pain, combined with either an ESR or CRP level and with or without a VAS GH (10). Joint examinations were performed for pain and swelling as recommended by the European League Against Rheumatism (EULAR) Standing Committee for International Clinical Studies Including Therapeutic Trials (ESCISIT) (18). GH was scored on a VAS from 0 (very well) to 100 (very poor). ESR was measured in mm/hour by the Westergren technique using a StaRRsed analyzer (Mechatronics LLC, Etten-Leur, The Netherlands). CRP level was measured in mg/dl by Tina-quant CRP (Roche Diagnostics, Mannheim, Germany).
A sensitivity analysis of the performance of the 4 DAS28 formulas in pregnant women was performed. For this purpose, we first determined the pregnancy-induced deviations of each component of the DAS28 observed in the healthy women. For each healthy woman, the third trimester and 12 weeks postpartum values of each component of the DAS28 were entered into the DAS28 formulas. Then the pregnancy-induced deviation in DAS28 for each component was calculated by subtracting the results of each woman separately at these different time points. Subsequently, the mean pregnancy-induced deviations in DAS28 were calculated for each component using the results of all healthy women. Finally, the mean pregnancy-induced deviations were entered into the 4 different DAS28 formulas to perform the sensitivity analysis.
Patients with RA were categorized according to their disease activity as determined by the 4 different DAS28 formulas in the third trimester of their pregnancy. This was rated according to the 4 criteria for categorizing disease activity as defined by ESCISIT (18). The categories are defined as high disease activity (DAS28 score >5.1), intermediate disease activity (3.2<DAS28≤5.1), low disease activity (2.6<DAS28≤3.2), and clinical remission (DAS28 score ≤2.6) (10).
The conventional HAQ score was determined using the validated Dutch translation of the Stanford HAQ, which considers the use of devices and aids (12, 13, 19). In a consensus meeting with researchers of our department, we defined the items and categories that might be most influenced by pregnancy.
To determine whether the assumptions of the consensus meeting were valid, we calculated the percentages of our healthy women scoring on the individual HAQ items. Based on the consensus and the rating of the individual HAQ items, modification of the conventional HAQ into 2 HAQ variants was recommended, both within the HAQ scoring rules (12). The first HAQ variant (HAQv1) was calculated on the basis of 6 rather than 8 categories, omitting arising and doing activities. The second HAQ variant (HAQv2) was calculated without selected items per category, i.e., dressing, getting in and out of bed, climbing 5 steps, getting in and out of a bath, bending down to get clothes from the floor, and getting in and out of a car. The 2 HAQ variants were calculated for patients with RA as well as for the healthy women. Finally, the performances of the conventional HAQ scores and the 2 HAQ variants were compared for all time points. The pregnancy-induced deviations of the conventional HAQ and of the 2 HAQ variants were calculated by the subtraction of the third trimester HAQ scores from that of 6 weeks postpartum.
Continuous variables were presented as the mean ± SD, i.e., VAS GH, ESR, and CRP level. Student's t-test was used to compare different time points of the continuous variables. Patients with RA were categorized in the 4 disease activity groups according the 4 different DAS28 formulas in the third trimester.
HAQ scores were presented as medians and interquartile ranges. The differences in HAQ scores between the first and third trimester and between conventional HAQ and HAQ variants were tested by Wilcoxon's nonparametric test. A difference in HAQ score of >0.25 between 2 time points was considered to be clinically relevant (12). A 2-sided P value ≤0.05 was considered statistically significant. SPSS for Windows, version 12.0 (SPSS, Chicago, IL) was used.
This study was in compliance with the Helsinki declaration, and the ethical committee at the Erasmus University Medical Center Rotterdam approved this study.
All 30 patients and 32 participating healthy women had successful pregnancies. As shown in Table 1, mean age, percentage of nulliparity, and parity were comparable between the healthy women and the patients with RA.
RA = rheumatoid arthritis; RF = rheumatoid factor; NS = not significant.
Age at first trimester, mean ± SD years
30 ± 3.7
32 ± 4.4
Nulliparous women, no. (%)
Number of previous pregnancies, mean ± SD
0.9 ± 1.0
1.0 ± 1.1
Disease duration, median (range) months
Erosions present, %
RF present, %
Pregnancy-induced deviations of components of the DAS28.
In the healthy women, the influence of pregnancy was different for each component of the DAS28. Pregnancy influenced the VAS GH to a small extent. The mean VAS GH of the healthy women was almost equal during pregnancy and postpartum (Figure 1). Despite this result, the mean pregnancy-induced deviation of the 2 possible DAS28 with 4 variables due to VAS GH was 0.22 in healthy women, calculated between the third trimester and 12 weeks postpartum.
Pregnancy influenced ESR considerably. The mean ESR of the healthy women differed statistically significantly between pregnancy and postpartum period (P = 0.03); ESR increased statistically significantly between the first and third trimesters (P = 0.01) from 10 mm/hour to 33 mm/hour. The calculations of the mean ESR at the third trimester excluded 1 woman who had sinusitis (ESR 120 mm/hour). The mean ESR returned to normal (<20 mm/hour) directly postpartum (Figure 1). The mean pregnancy-induced deviations of the DAS28 using ESR with 4 variables (DAS28-ESR-4) and the DAS28-ESR with 3 variables (DAS28-ESR-3) due to ESR were 1.07 and 1.16, respectively, in healthy women, regarding the different ESR levels at the third trimester and 12 weeks postpartum.
Pregnancy influenced CRP levels slightly. The mean CRP levels of the pregnant healthy women differed statistically significantly from those postpartum (P = 0.03). The mean CRP levels were equal between the first trimester (0.7 mg/dl) and the third trimester (1.0 mg/dl) (Figure 1). The calculations at the third trimester excluded 1 woman who had sinusitis (CRP level 7.6 mg/dl). The mean CRP level returned to normal (<0.7 mg/dl) directly postpartum. The mean pregnancy-induced deviations of the DAS28 using CRP level with 4 variables (DAS28-CRP-4) and the DAS28-CRP with 3 variables (DAS28-CRP-3) due to CRP were 0.25 and 0.27, respectively, in healthy women, calculated between the third trimester and 12 weeks postpartum.
The results of the sensitivity analysis of the 4 different DAS28 formulas differed considerably. The calculated mean pregnancy-induced deviations of the VAS GH, ESR, and CRP level for healthy women were added as in the 4 DAS28 formulas (see Appendix A). The total pregnancy-induced deviations per DAS28 were +1.16 for the DAS28-ESR-3, +1.29 for the DAS28-ESR-4, +0.27 for the DAS28-CRP-3, and +0.47 for the DAS28-CRP-4. Thus, healthy pregnancy influenced the DAS28-CRP-3 the least and the DAS28-ESR-4 the most.
When, according to the 4 DAS28 variants, patients in their third trimester were categorized in high, intermediate, and low disease activity and remission, large differences were found in the percentages of patients per category (Table 2). The percentages of clinical remission ranged between 0% and 23%. The highest total percentage (46%) of women with low disease activity or remission during the third trimester was identified by the DAS28-CRP-3, whereas according to the DAS28-ESR-3 and DAS28-ESR-4 the percentage was only 25%.
Table 2. Percentages of patients with rheumatoid arthritis expressing high, intermediate, and low disease activity or remission in the third trimester based on calculation of disease activity by 4 different DAS28 formulas*
High disease activity (DAS28 >5.1)
Intermediate disease activity (3.2<DAS28≤5.1)
Low disease activity (2.6<DAS28≤3.2)
Remission (DAS28 ≤2.6)
DAS28 = Disease Activity Score including 28-joint counts; DAS28-CRP-3 = DAS28 using C-reactive protein with 3 variables; DAS28-CRP-4 = DAS28 using C-reactive protein with 4 variables; DAS28-ESR-3 = DAS28 using erythrocyte sedimentation rate with 3 variables; DAS28-ESR-4 = DAS28 using erythrocyte sedimentation rate with 4 variables.
DAS28-CRP-3 (n = 30)
DAS28-CRP-4 (n = 30)
DAS28-ESR-3 (n = 28)
DAS28-ESR-4 (n = 28)
Pregnancy-induced deviations of the HAQ.
Because the healthy women answered more HAQ items “with some difficulty” or “with much difficulty,” they had difficulties in functioning during pregnancy. All items defined in the consensus meeting were marked with at least “some difficulty” in >20% of healthy women (Table 3). Only 1 item, i.e., walking outdoors, was not defined, but was also marked in >20% of healthy women in the third trimester.
Table 3. Percentages per item of the Health Assessment Questionnaire (HAQ) categories scored by healthy women in the first, second, and third trimesters and 6 weeks postpartum, showing the items that were marked as at least “with some difficulty” by 20% of the women in the third trimester*
HAQ category and item
6 weeks postpartum
“Unable to do” was never marked. The remaining percentage per item was ascribed to “no difficulty.”
Defined at the consensus meeting on the basis that they were likely to be most influenced by pregnancy.
HAQ 1.1: dress yourself, including tying shoelaces and doing buttons†
Among the categories marked most frequently during pregnancy were the categories as defined at the consensus meeting, i.e., arising (69%) and doing activities (59%). The category reaching, however, was also marked in 69% of healthy women in the third trimester. However, this third category could not be taken into account in a HAQ variant simultaneously with the other 2 categories, because calculating the HAQ score requires at least 6 categories.
Box plots demonstrated the differences of the HAQ and the 2 HAQ variant scores (HAQv1 and HAQv2) during pregnancy and postpartum (Figure 2). A third HAQ variant (not shown) was calculated on the basis of omitting items with the highest number of marks, instead of only the items considered in the consensus meeting. The scoring of this third HAQ variant did not perform better than the HAQv2.
The median pregnancy-induced deviation in HAQ score measured with the HAQ score between the third trimester and 6 weeks postpartum was 0.5. This pregnancy-induced deviation in HAQ score was clinically relevant and statistically significant. Although the use of the HAQv1 and HAQv2 reduced the pregnancy-induced deviation in HAQ scores between the third trimester and 6 weeks postpartum to 0.33 and 0.25, respectively, the influence of pregnancy on the scoring of the HAQ cannot totally be disregarded.
Median HAQ scores of the patients with RA increased statistically significantly by 0.31 from the first to the third trimester (P = 0.04) (Table 4). The median HAQv1 score also increased significantly by 0.42 (P = 0.04). Although the median HAQv2 score increased by 0.17 between the first and the second trimester, it decreased in the third trimester. The median HAQv2 scores between the first and third trimester, however, differed only 0.07 (P = 0.06).
Table 4. Health Assessment Questionnaire (HAQ) score and the 2 HAQ variants of women with rheumatoid arthritis during pregnancy*
1st trimester (n = 30)
2nd trimester (n = 30)
3rd trimester (n = 29)
6 weeks postpartum (n = 29)
Values are the median (interquartile range). HAQv1 = 1st HAQ variant scored without 2 categories (arising and activities); HAQv2 = 2nd HAQ variant scored without items 1.1, 2.2, 4.2, 5.2, 6.2, and 8.2.
P < 0.05 for HAQ score at 3rd trimester compared with HAQ score at 1st trimester.
Not significant for HAQv2 score at 3rd trimester compared with HAQv2 score at 1st trimester.
Our study demonstrates not only that pregnancy influences the scoring of the DAS28 and HAQ, but also that the HAQ variants can be used to reduce the influence of pregnancy on scoring the HAQ. The study also shows that the 2 RA assessment tools that performed best during pregnancy were the DAS28-CRP calculated without GH and the HAQ variant calculated without 6 selected items, although this variant did not totally preclude the influence of pregnancy.
Because the DAS28 had never previously been used in studies of RA in pregnancy, each component of the 4 different DAS28 formulas was first regarded separately in order to properly estimate the contribution pregnancy made to the formulas.
First, we showed that in healthy women, the mean GH score hardly changed during pregnancy, but was higher than postpartum. Although this difference was small, it still resulted in a mean pregnancy-induced deviation of 0.22 in both DAS28 formulas incorporating the GH. The GH during pregnancy in patients with RA had not been studied before; only patients' global assessment of RA had been studied. In 2004, Ostensen et al (8) prospectively studied patients' global assessments of RA in 10 pregnancies from preconception until 24 weeks postpartum. Patient assessment decreased by an average of 15 mm during pregnancy, and increased by 10 mm postpartum. The differences in GH measured in our 30 patients with RA during pregnancy and postpartum did not completely follow this pattern, because the mean GH of patients with RA decreased by only 1 mm during pregnancy and increased by 12 mm postpartum. During pregnancy, the GH and the patient's global assessment of RA may not be interchangeable, because the former scored health and disease activity together, whereas the latter scored only disease activity. The mean increases postpartum in the GH (12 mm) and in patients' global assessment of RA (10 mm) were comparable. This suggests that outside pregnancy both assessment tools solely scored the difference in disease activity of RA, which will have the greatest influence on the GH postpartum. Based on these findings, we suggest that, during pregnancy, DAS28 formulas can be better used without GH.
Second, in line with previous studies (14), our study demonstrates that ESR increases substantially during pregnancy. On the basis of earlier studies, 30% of patients with RA should be in clinical remission during pregnancy (4–6). The percentages of our patients with RA in clinical remission as determined by the DAS28-ESR formulas with or without GH were only 11% and 0%, respectively, at the third trimester. The DAS28-ESR will therefore certainly be unreliable, and therefore it is not a useful tool for assessing RA during pregnancy.
Third, during healthy pregnancy, a reference CRP level is lacking, especially because it is unknown whether CRP levels rise or fall. In our healthy women the levels measured were slightly higher during pregnancy, but the CRP levels did not increase as pregnancy proceeded. In gynecologists' daily practice, CRP level is already a useful marker of inflammation and infection, and is starting to be used in high-risk obstetrics such as gestational diabetes and preeclampsia (20). The higher CRP level we found during pregnancy agrees nicely with the slightly higher mean CRP levels (0.45 mg/dl) in early pregnancy (4 weeks of gestation) determined in women who became pregnant after in vitro fertilization procedure, and with the lower levels (0.13 mg/dl) in women in whom the procedure had failed (15). The higher CRP levels during pregnancy resulted in a pregnancy-induced deviation of the DAS28 of +0.25, which therefore performs better than that of the ESR (deviation of +1.1). The DAS28-CRP might therefore be a more reliable tool for assessing RA during pregnancy.
Last, although the joint scores of the DAS28 have never been designed for use in non-RA patients, it is conceivable that pregnancy potentially results in a joint score due to edema or to pain. Therefore, we performed joint counts in an additional 34 healthy pregnant women in different stages of pregnancy (7 in the first trimester, 10 in the second trimester, 17 in the third trimester). None of the examined women had swollen joints, and only 2 of them had 1 painful joint each. There was neither a correlation between edema and the presence of painful or swollen joints nor between the stage of pregnancy and the presence of painful or swollen joints. Therefore, we are convinced that the sensitivity analyses performed without joint scores but with ESR or CRP level, and with or without a GH are still valid.
In total, of the 4 DAS28 formulas, the DAS28-CRP calculated without GH was shown by the sensitivity analysis to be the least influenced by pregnancy itself. Therefore, this might be the best tool for scoring disease activity during pregnancy. Patients in their third trimester of pregnancy were categorized into different disease activity groups according to their DAS28-CRP score. DAS28-CRP calculated without GH showed the highest percentage of patients in clinical remission. The use of DAS28-CRP may still result in some overestimation of disease activity because CRP level is elevated during pregnancy compared with nonpregnancy, and therefore it may result in an underestimation of the percentage in clinical remission. Because a gold standard is not available to measure either the quantity of this underestimation or the percentage of misclassification that occurred, we suggest that pregnant patients with RA are best categorized according to their level of disease activity measured by DAS28-CRP without GH. This is important not only for research purposes but also for clinical practice in keeping an adequate control of disease activity.
The median HAQ score of our pregnant patients with RA was 1.0 at the third trimester, and is comparable with the median HAQ score of 0.9 calculated at the third trimester of 140 pregnant patients with inflammatory arthritis prospectively studied by Barrett et al from late pregnancy to postpartum (5). At the third trimester, the patients in the study by Barrett et al completed 2 HAQs, one pertaining to recalled disability before pregnancy and one to current disability. Their patients completed the HAQ at 4 and 26 weeks postpartum. However, their recalled disability before pregnancy was higher than all the HAQ scores prospectively measured. This pattern is very different from that of the HAQ we scored prospectively, which increased from the first trimester until the third trimester. The retrospective design of their study may have introduced recall bias. Postpartum, however, the median HAQ scores were again comparable: our patients had a median score of 0.75 and their patients a median score of 0.60 (5).
In contrast to what we found, Ostensen et al noted no difference at all between HAQ scores measured during pregnancy in 10 prospectively studied pregnancies of patients with RA (8). However, in that study the disease activity of the patients decreased at the same time and therefore one should have expected an increase in functionality. Because this was not the case, an explanation may be that pregnancy had also contributed to stable instead of lower HAQ scores.
Our results of the performance of the HAQ during pregnancy and results of former studies underline the arguments made by Nelson (16): the HAQ is not a good RA assessment tool during pregnancy. Therefore, we used different HAQ variants in an attempt to minimize the influence of healthy pregnancy on the HAQ. In healthy pregnant women, a pregnancy-induced increase in HAQ score was shown by all our variants that applied the HAQ scoring rules, such as the 2 defined in the consensus meeting, and those made on the basis of items with a high number of marks (12, 13). Despite this, the HAQ variants calculated without the 6 selected items generally performed best in this study. We have to conclude, however, that even this HAQ variant cannot preclude all influences of pregnancy on the assessment of functionality, and that each HAQ should therefore be used with great care during pregnancy.
In conclusion, pregnancy has various degrees of influence when calculating disease activity by 4 different DAS28 formulas and scoring functionality with the HAQ and HAQ variants. During pregnancy, the DAS28-CRP may therefore be the best tool for calculating disease activity and the percentage of patients with RA in clinical remission. The best tool for measuring functionality may be the HAQ variant from which 6 selected items were omitted.
Dr. de Man had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Hazes, de Man, Dolhain.
Acquisition of data. Hazes, de Man, van de Geijn, Krommenhoek, Dolhain.
Analysis and interpretation of data. Hazes, de Man, Dolhain.
Manuscript preparation. Hazes, de Man, van de Geijn, Krommenhoek, Dolhain.
Statistical analysis. Hazes, de Man, Dolhain.
We thank the participants of the PARA Study for their participation and the research nurses and assistants for their contribution to the data collection. We are grateful to Katinka Huender and her colleagues of the midwifery practice Voorburg for their cooperation.
t28 = number of tender joints of 28 joints examined for tenderness; sw28 = number of swollen joints of the 28 joints examined for swollenness; 28 joints = 28 tender and swollen joint scores include the same joints (shoulders, elbows, wrists, metacarpophalangeal joints, proximal interphalangeal joints, and the knees); ESR = Westergren's erythrocyte sedimentation rate (in mm/hour); CRP = C-reactive protein (in mg/liter [instead of in mg/dl as shown in Figure 1]); GH = global health rated on a 100-mm visual analog scale (GH was rated on a line from 0 [very well] to 100 [very poor]).