Abstract
Rationale: Studies of pregnancy in cystic fibrosis (CF) have shown no short-term harmful effects, but there are no long-term studies on the impact of motherhood.
Objectives: This study sought to evaluate longer-term physiologic and functional outcomes in women with CF reporting a pregnancy, with the intent of assessing how the demands of parenting impacted on disease course.
Methods: Using 1994 to 2005 Epidemiologic Study of Cystic Fibrosis data, we developed a propensity score to match women reporting a pregnancy at a 1:10 ratio with never-pregnant control subjects and compared clinical outcomes, health-related quality of life, and health care use.
Measurements and Main Results: One hundred nineteen pregnant women presumed to have become mothers were matched with 1,190 control subjects, a median of 6.0 years (range 1.8–11.1 yr) from the pregnancy. No differences were found in annualized change from baseline FEV1 and body mass index, in respiratory signs and symptoms, or in prescribed chronic therapies. Women who had been pregnant were treated for more pulmonary exacerbations and had more illness-related clinic visits but showed no increase in prescribed chronic therapies. They also reported lower health-related quality-of-life scores for Respiratory Symptoms, Physical Functioning, Vitality, and Health Perceptions.
Conclusions: Pregnancy and motherhood do not appear to accelerate disease progression but lead to more illness-related visits, pulmonary exacerbations, and a decrease in some domains of quality of life. These differences presumably reflect the impact of the physical and emotional challenges of early motherhood on disease self-management.
Improvements in the health and survival of individuals with cystic fibrosis (CF) have led to increasingly normal lifestyles, including successful pregnancies in women with CF. Studies published over the past 15 years have shown that pregnant women with CF receive more intensive monitoring and aggressive treatment during pregnancy and have an increased likelihood of receiving treatment for diabetes both during and after pregnancy. However, pregnancy is not associated with any immediate decline in health status or impact on survival (1–4).
On the other hand, little is known about the long-term physiologic and functional outcomes of women after pregnancy, when the demands of parenting might adversely impact a mother’s ability to devote adequate time and energy to disease self-management. We hypothesized that the experience of motherhood might lead to a decline in pulmonary health and emotional well-being even though the physical experience of pregnancy might not have a detrimental effect. To address this hypothesis, we compared long-term health care use, physiologic outcomes, and health-related quality of life (HRQoL) in women up to 11 years after pregnancy with those from a matched, never-pregnant cohort.
This analysis was conducted using data from the Epidemiologic Study of Cystic Fibrosis (ESCF), a prospective, multicenter, encounter-based observational study designed to characterize the natural history of CF in a large population of patients in the United States and Canada from 1994 to 2005 (5). Clinical, therapeutic, microbiologic, and pulmonary function data were collected throughout the study. Several additional variables (including those relevant to socioeconomic status and HRQoL) were collected beginning in 2003. The reported FEV1 values were converted to percent predicted using reference equations from Wang and colleagues (6) for female subjects through age 15 years and for male subjects through age 17 years, and Hankinson and colleagues (7) at older ages.
This analysis included women who were categorized based on their pregnancy status during the years 1994 to 2003 and compared clinical outcomes during the endpoint period 2004 to 2005. Pregnancy status was reported at each encounter in ESCF, but it was not recorded whether these pregnancies resulted in live births. Therefore, we included only those pregnancies that appeared to span at least 30 weeks. If a woman was reported to have had multiple pregnancies, the earliest was selected. Baseline clinical measurements were taken from the closest previous clinic visit that had occurred at least 60 (but not more than 365) days before the visit at which a pregnancy was first indicated. To be included in the analysis, women were required to have at least two encounters spanning 30 or more days and at least two FEV1 measures spanning 30 or more days during the 2004 to 2005 endpoint period.
These pregnant women were each matched at their baseline visit with 10 never-pregnant women using a propensity score (8) that included the following variables: age at baseline visit; CF genotype (homozygous Delta F508, heterozygous Delta F508, other, unknown); best FEV1 percent predicted in the 6 months before the baseline visit; age at diagnosis; Staphylococcus aureus and Haemophilus influenzae status (positive, negative, unknown) in the year before the baseline visit; intravenous antibiotic treatment for pulmonary exacerbation (yes/no; prorated count) in the year before the baseline visit; use of oral and inhaled bronchodilators, oral nutritional supplements, and pancreatic enzymes at the baseline visit; clinic visits (two or more visits in past year); and calendar year of baseline visit. Variables considered for inclusion in the propensity score but not included in the final model (P > 0.20) included race (white non-Hispanic, black non-Hispanic, Hispanic [any], other/unknown), diagnosis method (clinical symptoms, family history, screening), signs and symptoms (cough, sputum, clubbing, crackles, wheeze), other microbiology (Pseudomonas aeruginosa, Burkholderia cepacia, Stenotrophomonas maltophilia, methicillin-resistant S. aureus), indication of liver disease including elevated liver enzymes, use of other chronic therapies (oral and inhaled antibiotics, dornase alfa, mast cell stabilizers, oral and inhaled corticosteroids, enteral and parenteral nutritional supplements), measures of height and weight (body mass index [BMI], BMI z score, height-for-age z score, weight-for-age z score), other lung function measures (FVC and forced expiratory flow, midexpiratory phase), and counts of exacerbations and clinic visits.
For each pregnant woman, every nonpregnant woman was evaluated as a possible match at every possible baseline visit with adequate data. The most difficult-to-match women were matched first, and then that nonpregnant woman was eliminated from further consideration. This process continued until every pregnant woman had 10 matches. In addition to requiring the propensity scores to be within 0.01, additional matching criteria were applied: ages had to within 5 years, the baseline dates had to be within 730 days, and the exacerbation status (yes/no) had to be the same. Matching on propensity score was designed to minimize possible indication bias (9) arising from the likelihood that women who become pregnant tend to be healthier.
Pulmonary exacerbations and health care use during the endpoint period of 2004 to 2005 were compared using repeated-measures models controlling for the matching (i.e., these models accounted for the correlation among cases and their controls). Group differences in clinical status and treatment were analyzed in the same period using conditional logit models (controlling for the matching). HRQoL was measured using the Cystic Fibrosis Questionnaire-Revised (CFQ-R), a reliable and well-validated disease-specific instrument (10–12); group differences in mean scores were assessed with repeated-measures models controlling for the matching. All analyses were initially controlled for socioeconomic status measured by the median household income of the subjects’ postal code of residence according to 2000 U.S. census data (13). Inclusion of this variable made no substantive impact on effect estimates but reduced the number of subjects who could be included in the analysis due to missing data; therefore, we report the results of models that do not adjust for socioeconomic status in the endpoint period.
The inclusion criteria for both groups in our primary analysis included the availability of follow-up data, so differential loss to follow-up between pregnant and nonpregnant women, due to differences in mortality or any other reason, could have biased the assessment of outcomes during the endpoint period. To evaluate this possibility, we created a different pregnancy cohort and matched control group from the ESCF dataset using the same definitions and matching algorithm as the original cohort with two differences: (1) women were not required to have data in the endpoint period, and (2) control women were required to have a minimum of 9 months of data after their baseline visit to be more comparable to pregnant women (for whom the minimum follow-up was around 9 mo). The death rate analysis was performed using the subset of women whom we could link with the CF Foundation Patient Registry (14), which provided a more consistent and comprehensive source of information for date of death. The number lost to follow-up and the number of deaths that had occurred by the end of 2005 in each group were compared using the Chi square test statistic and Kaplan-Meier curves.
All analyses were performed using SAS Version 9.1 (Cary, NC). The study was approved by the Copernicus Group institutional review board (tracking number OVA1-03-008). Where required, it was approved by local institutional review boards, and participants or their guardians provided informed consent.
In total, 119 women met the pregnancy inclusion criteria and were compared with a matched control group of 1,190 never-pregnant women. The median time between pregnancy and the endpoint period was 6.0 years (range, 1.8–11.1 yr). There were no significant differences between the groups in demographic characteristics (Table 1), indicating an appropriate matching methodology, except that the pregnant group was on average 2 years older than the never-pregnant group. Furthermore, there were no statistically significant group differences in clinical characteristics at baseline (Table 2).
| Characteristic | Pregnant (n = 119) | Never Pregnant (n = 1,190) |
|---|---|---|
| Age at baseline,* mean ± SD, yr | 23.4 ± 4.8 | 21.6 ± 6.5 |
| Race, n (%) | ||
| White, non-Hispanic | 112 (94.1) | 1,125 (94.5) |
| Black, non-Hispanic | 5 (4.2) | 26 (2.2) |
| Hispanic (any) | 1 (0.8) | 32 (2.7) |
| Other/mixed/unknown | 1 (0.8) | 7 (0.6) |
| Genotype, n (%) | ||
| deltaF508/deltaF508 | 45 (37.8) | 459 (38.6) |
| deltaF508/other | 43 (36.1) | 386 (32.4) |
| Other | 14 (11.8) | 98 (8.2) |
| Unknown | 17 (14.3) | 247 (20.8) |
| Median family income by postal code, n (%) | ||
| <$40K | 25 (21.0) | 196 (16.5) |
| $40K to < $50K | 32 (26.9) | 310 (26.1) |
| $50K to < $60K | 29 (24.4) | 238 (20.0) |
| ≥$60K | 30 (25.2) | 377 (31.7) |
| Unknown | 3 (2.5) | 69 (5.8) |
| Characteristic | Pregnant (n = 119) | Never Pregnant (n = 1,190) |
|---|---|---|
| FEV1, % predicted, mean ± SD | 74.1 ± 19.6 | 75.5 ± 22.0 |
| BMI z score, mean ± SD | −0.23 ± 0.83 | −0.31 ± 0.99 |
| Weight-for-age z score, mean ± SD | −0.42 ± 0.88 | −0.55 ± 1.08 |
| Crackles, n (%) | ||
| Yes | 44 (37.0) | 349 (29.3) |
| No | 75 (63.0) | 841 (70.7) |
| Daily sputum, n (%) | ||
| Yes | 67 (56.3) | 600 (50.8) |
| No | 52 (43.7) | 582 (49.2) |
| Daily cough, n (%) | ||
| Yes | 83 (69.8) | 797 (67.4) |
| No | 36 (30.3) | 386 (32.6) |
| Wheezing, n (%) | ||
| Yes | 10 (8.4) | 82 (6.9) |
| No | 109 (91.6) | 1,108 (93.1) |
| Clubbing, n (%) | ||
| Yes | 77 (64.7) | 777 (65.3) |
| No | 42 (35.3) | 413 (34.7) |
| Diabetes, n (%) | ||
| Yes | 14 (11.8) | 149 (12.5) |
| No | 105 (88.2) | 1,041 (87.5) |
We compared the annualized change in FEV1 and BMI between the baseline value and the best value during the 2004 to 2005 endpoint period and found no significant group differences (Table 3). The results were not substantively different when these outcomes were compared using change from the baseline value to each (rather than best) value reported during the 2004 to 2005 endpoint period. We found a trend toward increased cough, sputum, and crackles in women after pregnancy, but the differences were not statistically significant (Table 4). Routine maintenance therapies and presence of diabetes (as indicated by use of insulin or oral hypoglycemic agents) were also similar between groups. There was one exception: among women who were receiving inhaled corticosteroids at baseline, women who had been pregnant were less likely than control subjects to be receiving them during the endpoint period (74.3 vs. 92.6%, P = 0.007).
| Outcome | Pregnant | Never Pregnant | Mean Difference | P Value | ||
|---|---|---|---|---|---|---|
| n | Mean ± SE | n | Mean ± SE | Mean ± SE | ||
| FEV1, % predicted | 119 | −1.54 ± 0.34 | 1,190 | −1.08 ± 0.13 | −0.46 ± 0.35 | 0.19 |
| BMI z score | 117 | 0.06 ± 0.02 | 1,161 | 0.09 ± 0.01 | −0.03 ± 0.02 | 0.22 |
| Weight z score | 119 | 0.05 ± 0.02 | 1,187 | 0.08 ± 0.01 | −0.03 ± 0.02 | 0.18 |
| Outcome* | BL | Pregnant | Never Pregnant | P Value† | P Value‡ | ||
|---|---|---|---|---|---|---|---|
| n | Percentage (95% CI) | n | Percentage (95% CI) | ||||
| Crackles | Yes | 40 | 72.7 (56.5–84.5) | 138 | 61.9 (52.3–70.6) | 0.22 | 0.063 |
| No | 74 | 37.6 (26.9–49.6) | 531 | 29.5 (25.0–34.4) | 0.16 | ||
| Daily sputum | Yes | 65 | 84.8 (73.7–91.8) | 352 | 79.7 (74.5–84.1) | 0.34 | 0.060 |
| No | 52 | 67.7 (53.3–79.4) | 269 | 52.8 (45.6–60.0) | 0.054 | ||
| Daily cough | Yes | 83 | 90.6 (82.0–95.3) | 563 | 85.8 (82.3–88.8) | 0.24 | 0.075 |
| No | 36 | 80.9 (64.4–90.8) | 124 | 66.6 (56.5–75.3) | 0.11 | ||
| Wheezing§ | No | 109 | 7.3 (3.7–14.0) | 1,010 | 4.7 (3.5–6.2) | 0.22 | N/A |
| Clubbing | Yes | 77 | 90.9 (82.1–95.6) | 491 | 89.0 (85.8–91.6) | 0.61 | 0.42 |
| No | 41 | 68.3 (52.6–80.8) | 137 | 61.3 (52.5–69.4) | 0.42 | ||
There were some substantive differences between the women who had been pregnant and their matched controls during the endpoint period. The pregnant group had more courses of intravenous antibiotics and more clinic visits; the difference in clinic visits was due to a greater number of sick visits (Figure 1). Of 260 women with available data regarding HRQoL (54 pregnant and 206 never pregnant), the pregnant group had significantly lower scores on the CFQ-R on the following scales: Physical Functioning, Vitality, Health Perceptions, and Respiratory Symptoms. There were no differences on the other CFQ-R scales: Role, Emotional, and Social Function, Body Image, Eating Disturbances, Treatment Burden, Weight, and Digestive Symptoms (Figure 2). The CFQ-R was first introduced into ESCF in 2003, so there were no baseline scores for comparison.
Figure 1. Health care use during the endpoint period 2004 to 2005. Comparison of women who had reported a pregnancy to those who had not. IV = intravenous.
[More] [Minimize]
Figure 2. Scores on the Cystic Fibrosis Questionnaire-Revised during the endpoint period 2004 to 2005. Comparison of women who had reported a pregnancy to those who had not.
[More] [Minimize]A Kaplan-Meier curve evaluating loss to long-term follow-up in ESCF showed a somewhat more rapid dropout rate in the nonpregnant control group compared with the women who had been pregnant (P < 0.001) (Figure 3). A comparison of survival using mortality data from the CF Foundation Patient Registry showed a nonsignificant trend toward greater mortality in the control group: 18% (41 of 228) of women who had been pregnant and 23% (522 of 2,280) of control subjects died before the end of the study period (P = 0.090) (Figure 4).
Figure 3. Kaplan-Meier plot showing loss to follow-up in the Epidemiologic Study of Cystic Fibrosis up to 2005. Comparison of women who had reported a pregnancy to those who had not.
[More] [Minimize]
Figure 4. Kaplan-Meier plot comparing survival of the Cystic Fibrosis Foundation Registry cohort up to 2005. Comparison of women who had reported a pregnancy between 1994 and 2003 to those who did not.
[More] [Minimize]Improved health outcomes in patients with CF have led to an increasing number of affected women who are having children. Several recent studies have shown that pregnancy in CF is not associated with an immediate decline in health status or survival, although it does lead to an increased use of chronic therapies and incidence of diabetes (1–4). We speculated that the intense demands of early parenting might impinge on a new mother’s time for disease self-management and therefore lead to worsening of long-term physiologic and functional outcomes. We found that, when compared with control subjects, women who had experienced a pregnancy a median of 6 years earlier had:
| • | Similar changes in lung function, nutritional status, respiratory signs and symptoms, and no difference in mortality; | ||||
| • | Similar use of routine outpatient clinic visits and maintenance therapies (except for a decrease in inhaled corticosteroids); | ||||
| • | More courses of intravenous antibiotics and illness-related outpatient visits; | ||||
| • | Lower HRQoL scores for Physical Functioning, Vitality, Health Perceptions, and Respiratory Symptoms; but no differences in Role, Emotional, and Social Function, Body Image, Eating Disturbances, Treatment Burden, Weight, and Digestive Symptoms. | ||||
From this analysis, it appears that motherhood has no more detrimental effect on the health of women with CF than does pregnancy itself. We found no significant deterioration in pulmonary function or nutrition during the years after pregnancy and no increase in the burden of treatment with chronic therapies. Nonetheless, these women had more pulmonary exacerbations and sick clinic visits. We conjecture that this could be due to a decrease in adherence to their home treatment regimen, which could eventually lead to increased morbidity. It may be noted that the point estimates of change in FEV1 and respiratory signs and symptoms were greater in the pregnant group; one could speculate that longer follow-up or a larger sample of pregnant women could have provided greater power and statistical significance to these observations, but no hint of an adverse effect on mortality was seen. The lack of a difference in the number of routine clinic visits went against our hypothesis that new mothers might have a difficult time keeping up with clinic appointments and follow-up. Differences in patient-reported experiences of respiratory symptoms, physical functioning, vitality, and health perceptions presumably reflect the impact of the physical and psychological demands of parenting added to the burden of an arduous CF treatment regimen. To our knowledge, there have been no studies evaluating the impact of pregnancy and motherhood on HRQoL in the general population with which we can compare our findings in women with CF (15).
One limitation of this study was our inability to confirm live birth or child-rearing arrangements. Although we do not know the exact proportion of our pregnant or control women who are parenting children, it is safe to assume that it is much higher in the pregnant group. Therefore, differences in the groups should be an indication of the impact of motherhood. In addition, we had no information on the number of children in the home (in either group), nor are there any data on breastfeeding, which might represent an added nutritional burden in the first year after pregnancy.
This observational study has other inherent limitations. Women with CF who become pregnant generally have better pulmonary function than women who do not become pregnant (4), but they may also differ in other ways. This makes our analysis subject to indication bias, also known as confounding by indication (9). Of several possible approaches that can be used in an attempt to avoid this pitfall (16), we used propensity score matching, which reduces bias but does not adjust for unmeasured characteristics (17, 18). However, we believe that any indication bias that may exist in our population would also be found in future populations of women with CF who become pregnant, making our findings generalizable and externally valid.
Our analytic approach, which required the availability of data during the follow-up period, may have excluded women who died prematurely during or after their pregnancy, biasing the analysis toward a more favorable conclusion regarding the long-term consequences of pregnancy and motherhood. Comparisons of overall loss to follow-up and mortality are reassuring in this regard; there was a greater loss to followup in ESCF among women who had never been pregnant and no suggestion that there was any adverse effect on mortality from motherhood.
In summary, the findings of our analysis provide further reassurance to women with CF who wish to become pregnant. Just as earlier studies showed no short-term health consequences of pregnancy other than an increased incidence of new-onset diabetes, the current study found no long-term detrimental effect on pulmonary or nutritional outcomes and no substantial impact on the use of chronic therapies, although there was an increase in the number of sick clinic visits and courses of intravenous antibiotics. Certain aspects of HRQoL decreased, which may reflect the additional burden of responsibilities associated with motherhood in general. Our findings may guide the clinician in counseling women with CF who are contemplating pregnancy. An awareness of the greater need for illness-related outpatient visits and intravenous antibiotics and the impact of parenthood on HRQoL may help both patients and clinicians anticipate potential difficulties and implement mitigating strategies in women with CF who become mothers.
| 1 . | Gilljam M, Antoniou M, Shin J, Dupuis A, Corey M, Tullis DE. Pregnancy in cystic fibrosis: fetal and maternal outcome. Chest 2000;118:85–91. |
| 2 . | Goss CH, Rubenfeld GD, Otto K, Aitken ML. The effect of pregnancy on survival in women with cystic fibrosis. Chest 2003;124:1460–1468. |
| 3 . | Hardin DS, Rice J, Cohen RC, Ellis KJ, Nick JA. The metabolic effects of pregnancy in cystic fibrosis. Obstet Gynecol 2005;106:367–375. Epub 2005/08/02. |
| 4 . | McMullen AH, Pasta DJ, Frederick PD, Konstan MW, Morgan WJ, Schechter MS, Wagener JS. Impact of pregnancy on women with cystic fibrosis. Chest 2006;129:706–711. |
| 5 . | Morgan WJ, Butler SM, Johnson CA, Colin AA, FitzSimmons SC, Geller DE, Konstan MW, Light MJ, Rabin HR, Regelmann WE, et al. Epidemiologic study of cystic fibrosis: design and implementation of a prospective, multicenter, observational study of patients with cystic fibrosis in the U.S. and Canada. Pediatr Pulmonol 1999;28:231–241. |
| 6 . | Wang X, Dockery DW, Wypij D, Fay ME, Ferris BG Jr. Pulmonary function between 6 and 18 years of age. Pediatr Pulmonol 1993;15:75–88. |
| 7 . | Hankinson JL, Odencrantz JR, Fedan KB. Spirometric reference values from a sample of the general US population. Am J Respir Crit Care Med 1999;159:179–187. |
| 8 . | Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med 1997;127:757–763. |
| 9 . | Rothman KJ, Wentworth CE, 3rd. Mortality of cystic fibrosis patients treated with tobramycin solution for inhalation. Epidemiology 2003;14:55–59. |
| 10 . | Quittner AL, Buu A, Messer MA, Modi AC, Watrous M. Development and validation of The Cystic Fibrosis Questionnaire in the United States: a health-related quality-of-life measure for cystic fibrosis. Chest 2005;128:2347–2354. |
| 11 . | Quittner AL, Schechter MS, Rasouliyan L, Haselkorn T, Pasta DJ, Wagener JS. Impact of socioeconomic status, race, and ethnicity on quality of life in patients with cystic fibrosis in the United States. Chest 2010;137:642–650. |
| 12 . | Quittner AL, Sawicki GS, McMullen A, Rasouliyan L, Pasta DJ, Yegin A, Konstan MW. Psychometric evaluation of the Cystic Fibrosis Questionnaire-Revised in a national sample. Qual Life Res 2012;21:1279–1290. |
| 13 . | Schechter MS, McColley SA, Silva S, Haselkorn T, Konstan MW, Wagener JS. Association of socioeconomic status with the use of chronic therapies and healthcare utilization in children with cystic fibrosis. J Pediatr 2009;155:634–639. |
| 14 . | Cystic Fibrosis Foundation. Cystic Fibrosis Foundation Patient Registry 2009 Annual Data Report. Bethesda, Maryland: 2010. |
| 15 . | Coyle SB. Health-related quality of life of mothers: a review of the research. Health Care Women Int. 2009;30:484–506. |
| 16 . | Haro JM, Kontodimas S, Negrin MA, Ratcliffe M, Suarez D, Windmeijer F. Methodological aspects in the assessment of treatment effects in observational health outcomes studies. Appl Health Econ Health Policy 2006;5:11–25. |
| 17 . | Glynn RJ, Schneeweiss S, Sturmer T. Indications for propensity scores and review of their use in pharmacoepidemiology. Basic Clin Pharmacol Toxicol 2006;98:253–259. |
| 18 . | Sturmer T, Rothman KJ, Avorn J, Glynn RJ. Treatment effects in the presence of unmeasured confounding: dealing with observations in the tails of the propensity score distribution–a simulation study. Am J Epidemiol 2010;172:843–854. |
Supported by Genentech, which provided funding and support for data collected by the Epidemiologic Study of Cystic Fibrosis, honoraria to the authors for service as members of the Scientific Advisory Group, and payment for statistical analysis.
Author Contributions: M.S.S., A.L.Q., M.W.K., S.J.M., D.J.P., and A.M. were involved in the conception, hypotheses delineation, and design of the study. M.S.S., A.L.Q., M.W.K., S.J.M., D.J.P., and A.M. were involved in the acquisition of the data or the analysis and interpretation of such information. M.S.S., A.L.Q., M.W.K., S.J.M., and D.J.P. were involved in writing the article or substantial involvement in its revision prior to submission.
Genentech played no role regarding study design of this analysis, the analysis itself, the writing of the manuscript, or the decision to submit the manuscript for publication.
Author disclosures are available with the text of this article at www.atsjournals.org.
