Health-related quality of life in women with nausea and vomiting of pregnancy: the importance of psychosocial context

Abstract

Objective

To test whether women with hyperemesis gravidarum (HG) demonstrated lower health-related quality of life (HRQoL) scores compared with those with nausea and vomiting of pregnancy (NVP).

Study Design

Women with HG or NVP were examined during the first trimester. Multivariate models identified characteristics of women at risk for low HRQoL, as measured by an NVP-specific HRQoL test and a generic HRQoL test, the Short Form (SF)-36.

Result

Although the SF-36 assessment did not discriminate between the two groups, the NVP-specific test showed that women with HG (N¼29) were 3–6 times more likely than women with NVP (N¼48) to have low HRQoL. Both tests demonstrated that perceived physical symptoms and multiple psychosocial factors, such as depression and marital status, seemed to be equally or more important than having HG.

Conclusion

Although a low HRQoL was associated with an HG diagnosis, multiple physical symptoms and psychosocial factors placed both groups of women at risk. Journal of Perinatology (2011) 31, 10–20; doi:10.1038/jp.2010.54; published online 22 April 2010

Introduction

Health-related quality of life (HRQoL) is defined as an individual’s functional experience of disease and treatment-related symptoms.¹ A woman’s perspective of her ability to maintain her expected level of function for personal, household and social responsibilities while confronting illness is now recognized as an important dimension of the quality of medical care.²

Existing literature suggests that women with uncomplicated pregnancies experience functional changes that can alter their ability to carry out their usual roles and diminish their quality of life,^3,4 and this is likely reflective of the physical demands of pregnancy on the body.⁵ Moreover, the decreased HRQoL experienced routinely by pregnant women may, in part, be attributed to the impact of nausea and vomiting of pregnancy (NVP),^6–9 which affects 70–85% of pregnant women.¹⁰ Nausea and vomiting of pregnancy not only affects the physical health of pregnant women, but can also negatively impact their family, social and occupational functioning.^11,12 Furthermore, the degree of this negative impact seems to be associated with the severity of NVP.^8,13,14

Severe and intractable NVP, hyperemesis gravidarum (HG), occurs in approximately 0.5–2% of pregnancies,¹⁵ and studies specific to HG have also detailed the significant physical, psychological, social and economic toll.^16,17 An ambiguity remains with regard to whether NVP and HG are independent conditions or a part of a continuum of illness, and the causes of both conditions remain unknown.¹⁸

One application of HRQoL measures is their utility in estimating the burden of illness. Understanding the health burden, that is, ‘the total impediment in physical, mental, and social functioning and well-being in the personal evaluation of health,’¹⁹ of NVP and HG is central to the interdisciplinary facets of perinatal clinical care, research and policy. Ware’s²⁰ conceptual model of HRQoL recognizes that a given health condition, in this case NVP, causes both physical and mental symptoms, and that the impact of these symptoms on HRQoL must be viewed within a social, or participatory, context. Thus, although physical symptom severity may have an important role in determining HRQoL, a woman’s social and functional requirements and her level of social support will contribute to the assessment of her health burden.

Both generic and disease-specific HRQoL instruments have been developed to facilitate the assessment of physical and psychosocial components of patients’ well-being and functioning using their self-reports.²¹ Taking advantage of both generic and disease-specific tools may help us to better understand the impact of NVP on HRQoL.²² Two tools lend themselves to the assessment of HRQoL in women with NVP: a well-established generic instrument, the Medical Outcomes Study Short Form (SF36),¹⁹ or its abbreviated version, the SF12, has been used three times to our knowledge to study NVP populations 6–8 and once in a self-reported HG population.⁹ A condition-specific instrument for NVP HRQoL assessment (NVPQOL) was developed by Magee et al.,²³ and was further validated by Lacasse and Berard.²⁴

It has been established that pregnancy alone is associated with lower SF36 scores in the physical functioning, physical limitations and bodily pain domains when uncomplicated pregnant women were compared with nonpregnant women.4 Among pregnant women, women with NVP seem to have lower SF36 scores when compared with asymptomatic pregnant women,⁹ and women with severe or moderate NVP have lower NVPQoL scores when compared with those with mild NVP.⁸ No study has, to the best of our knowledge, yet used either test directly to compare the HRQoL scores of women with HG vs severe or moderate NVP, or even to discriminate between women with severe vs moderate NVP.

In this study, in an effort to establish the importance of physical symptom severity of women with HG on HRQoL, we tested whether women with HG demonstrated lower HRQoL scores compared with those with NVP. We also identified and quantified the independent contribution of a variety of physical and psychosocial factors to HRQoL with respect to the presence of HG.

Methods

This is an exploratory study designed to evaluate HRQoL in women with HG and NVP in the first trimester of pregnancy. It is part of a larger comprehensive evaluation of the clinical and physiological parameters associated with NVP, and specifically with HG. The protocol was approved by the Institutional Review Board of the Health Sciences Campus of the University of Southern California. All participants were recruited for the study during the first trimester of pregnancy. Women with HG were recruited during admission to local hospitals because of the rarity of the condition. Local obstetricians and perinatologists treating HG patients were made aware of the study, and communicated with the study team when HG patients were admitted. Hyperemesis gravidarum was defined as persistent vomiting not related to other causes that required hospitalization for treatment with intravenous fluids or parenteral nutrition. Women with HG were evaluated after they were discharged from their hospitalization and were sufficiently recovered to undergo testing.

Women without HG were largely recruited from local clinics associated with the hospitals from which HG patients were recruited, and by advertisements in local media. Women were eligible for the study if they presented in the first trimester and had no condition responsible for nausea and vomiting symptoms other than pregnancy. Women without HG were divided into those with and without NVP on the basis of their answers given to four questions in the NVPQOL test regarding current symptoms.

This study design thus yielded three study groups: women with HG, women with NVP and asymptomatic women. In addition to completing the NVPQOL and the SF36 assessments, the women provided demographic and health history information, and were tested using the following instruments:

• The Symptoms Checklist 90 (SCL90) is a widely used self-report symptom inventory that is a measure of current psychological symptom status over the past week.²⁵ The SCL90 uses norm-based scoring methods, with higher scores associated with increased symptomatology. Several domains were examined, including depression, anxiety, somatic symptoms and a global measure.

• The Beck Depression Inventory-II²⁶ is a self-report symptom inventory consisting of 21 items that reflect symptoms over the past two weeks. The scale for each item ranges from 0 to 3, with 3 being the most symptomatic. The Beck Depression Inventory-II includes a cognitive/affective subscale and a somatic subscale, which were both included in this study.

• The Dyadic Adjustment Scale (DAS) is a commonly used instrument to measure the quality of adjustment to marriage and similar dyadic relationships.²⁷ The DAS has four domains: consensus, satisfaction, affectional expression and cohesion. Several items in the DAS were also examined independently, including agreement regarding household tasks (question 13), quarreling (question 21) and getting on each others’ nerves (question 22). The DAS uses norm-based scoring methods, and higher DAS scores were associated with better quality of the relationship.

• The Pregnancy Unique Quantification of Emesis (PUQE) instrument quantifies the severity of NVP.²⁸ Scores for each of three items ranged from 1 to 5, with 5 being the most severely symptomatic. Question 1 described nausea, question 2 described vomiting and question 3 described retching within the last 12 hours.

• The Marin Acculturation Scale for Hispanic subjects uses a Likert scale ranging from 1 to 5, with higher scores representing higher acculturation in the following three domains: personal use of the English language, listening to English media and use of English in social situations.²⁹

The potential physical, mental and social drivers of HRQoL for women with NVP were derived from the above tests. In addition, a history of motion sickness within the last 10 years was derived from a previous study by Mirabile.³⁰ There is some published evidence that women with a history of motion sickness may be more susceptible to NVP and HG,³¹ and that they may be more likely to vomit during pregnancy.³² The volume of a sample of unstimulated saliva was also measured to document the baseline activity of the parotid glands. The amount of saliva expectorated into a collection tube should yield an objective measure of the degree to which a patient’s salivation system has been activated.³³

Health-related quality of life outcomes were measured with the NVPQOL and SF36 assessments. The NVPQOL is a specific instrument designed to evaluate HRQoL in women with NVP within the last week.²³ It has a total of 30 items describing symptoms, each scored from1 all of the time to⁷ none of the time, and yields a total score (when averaged, ranging from 1 to 7, with 7 being the highest quality of life) and similar scores in four domains (physical symptoms, fatigue, emotions and social limitations). Although it has been validated,²⁴ it does not have published normative values. The SF36 (standard form) is a generic HRQoL measurement with demonstrated reliability and validity in a healthy population and across diverse patient groups.¹⁹ It focuses on both physical and mental wellbeing as perceived by the patient over the past 4 weeks. It provides two summary scores, physical and mental, and is comprised of eight domains: Physical Functioning, Role Physical, Bodily Pain, General Health, Vitality, Social Function, Role Emotional and Mental Health. It has a total of 36 items, and is scored using population-based norms.

After initial calculation of HRQoL, and comparison with population-based norms (in the case of the SF36 test), potential drivers of HRQoL and HRQoL outcomes were examined separately in the NVP and HG populations to determine whether they differed in the two groups. Asymptomatic women were not included in the multivariate analyses. As our results demonstrated that the potential drivers of HRQoL were very similar for both NVP and HG populations, both groups were combined for the sole purpose of identifying likely candidates to be used in multivariate modeling of HRQoL outcomes.

For each HRQoL domain, recursive partitioning algorithms (RPAs) were then used to examine potential interactions among the clinical factors, and to ascertain which women were most at risk for low HRQoL. Recursive partitioning is a nonparametric technique that produces a classification or decision tree in which subjects are assigned to mutually exclusive subsets according to a set of predictor variables.^34,35 It identifies subject subgroups with varying risks and may uncover interactions between predictors that may be overlooked in the traditional application of logistic regression to data sets with numerous predictors, as in our case. The decision tree was constructed by splitting subsets of the data set using all predictor variables to create two or more child nodes repeatedly, beginning with the entire data set. The predictor having the highest association with the target variable was selected for splitting. The association between each predictor variable and the target was computed using the analysis of variance F-test (for ordinal and continuous predictors) or Pearson’s X²-square test (for nominal predictors).

Multivariable linear regression models incorporating predictors and their salient interactions were then identified for each HRQoL outcome, and these results were combined with RPA models to produce the most parsimonious multivariate logistic regression models for each outcome to determine the odds ratios (ORs) and 95% confidence intervals for these predictors. Hyperemesis gravidarum was forced as a predictor into all models. Characteristics of women with and without HG at highest risk for poor HRQoL outcomes are described.

Categorical data were examined using X²-square testing, and continuous data were examined using nonparametric testing as appropriate (the Kruskal–Wallis test). Means are expressed with ± s.d. values, and P<0.05 was considered statistically significant. No adjustment was made for multiple comparisons because of the exploratory nature of the study and the resulting limitations to causal inference.36 All calculations were performed in SAS (v. 9.1, Cary, NC, USA), except for RPA models, calculations for which were performed in SPSS with the AnswerTree module (SPSS, Chicago, IL, USA).

Results

Of the 96 women enrolled into the study, 93 had complete data regarding NVPQOL: 29 had HG, 48 had NVP and 16 were asymptomatic. Owing to a delay in implementing the administration of SF36 into the study protocol, it was taken by fewer women (N=61): 19 with HG, 34 with NVP and 8 without symptoms. For the analyses in which asymptomatic women were eliminated, the denominators for the NVPQOL and SF36 analyses were 77 and 53, respectively.

Demographic characteristics did not differ across the groups. The mean maternal age was 27.6±6.1 years, and the mean gestational age at evaluation was 11.9±3.5 weeks. The majority of women were married or living with a partner (72.0%), did not attend college (61.3%) and had government health insurance (75.3%). Most were of Hispanic ethnicity (73.1%) and half were multiparous (52.7%). Only one patient in the entire study population reported ever receiving medication to treat depression before becoming pregnant, and none reported any medical treatment for postpartum depression.

Table 1 describes the HRQoL scores for women with and without HG for each of the four domains of the NVPQOL test. Asymptomatic women were not considered in this table because NVPQOL was used to separate women into asymptomatic and NVP groups. In spite of the small sample size, the NVPQOL seemed to discriminate between the NVP and HG groups in both the total score and the Emotions domain.

Table 1.

Health related quality of life scores for women with NVP vs women with HG for each of the domains of the NVPQOL assessment

Domain	Women with NVP N=48 median, Range	Women with HG N=29 median, range	P-value
Physical symptoms	3.44, 1.33–5.67	2.33, 1.13–6.89	0.0854
Fatigue	3.00, 1.00–6.75	2.25, 1.00–7.00	0.0599
Emotions	3.50, 1.86–7.00	3.00, 1.43–6.14	0.0352
Limitations	3.40, 1.00–6.20	2.60, 1.00–7.00	0.0720
Total	3.32, 1.86–5.78	2.64, 1.27–6.73	0.0337

Abbreviations: HG, hyperemesis gravidarum; NVP, nausea and vomiting of pregnancy; NVPQOL, Nausea and vomiting in pregnancy health related quality of life.

Possible scores range from 1 to 7, with 7 being the highest quality of life.

Table 2 describes the mean HRQoL scores for asymptomatic women, as well as for women with NVP and for women with HG for each of the eight domains and the two summary measures of the SF36. In five of the eight domains, and in the physical summary score, the NVP and HG groups seemed to have lower HRQoL scores when compared with the asymptomatic group of pregnant women. However, the SF36 did not seem to discriminate between the NVP and HG groups (Table 2), and statistical testing between the scores for these groups alone was unremarkable (data not shown). The SF36 scores for the NVP and HG groups were also tested against national norms for females of all ages, and the majority of scores were significantly lower (Table 2).

Table 2.

Mean health related quality of life scores for women with NVP vs women with HG for each of the domains of the medical outcomes study short form (SF36)

Domain	Score for Asymptomatic group (N=8) median and range	Score for NVP group (N=34) median and range	Score for HG group (N=19) median and range
Physical function	41.2, 25.5–50.7a	36.0, 14.9–50.7a	3.9, 19.2–50.7a
Role physicalb	53.2, 27.5–56.9	34.8, 17.7–56.9a,c	29.9, 17.1–56.9a,c
Bodily painb	60.0, 51.1–62.1a	41.8, 24.9–62.1c	37.4, 24.9–62.1c
General healthb	52.9, 48.2–63.9	45.3, 281.–61.5c	43.4, 25.8–60.1c
Vitalityb	55.2, 36.5–70.8	36.5, 20.9–64.6a,c	39.6, 20.9–64.6a,c
Social functionb	48.7, 24.1–56.9	35.0, 13.2–56.8a,c	29.6, 13.2–56.8a,c
Role emotional	32.6, 21.7–46.5a	32.6, 9.2–46.5a	27.9, 15.5–46.5a
Mental health	48.6, 30.3–61.3	37.3, 16.2–64.1a	41.6, 24.7–64.1
Physical summaryb	49.9, 47.7–67.9	42.0, 27.6–52.0a,c	35.8, 24.3–55.2a,c
Mental summary	45.4, 23.2–58.4	38.2, 14.6–57.2a	36.6, 19.3–62.1a

Abbreviations: HG, hyperemesis gravidarum; NVP, nausea and vomiting of pregnancy.

No statistically significant differences were found in the comparison between the NVP and HG groups.

^aIndicates that the P-value for a Mann–Whitney test comparing the indicated group with norms for US females (N=4032) was <0.05.

^bIndicates that the P-value for a Kruskal–Wallis test comparing all three groups was <0.05.

^cIndicates that the P-value for a Mann–Whitney test comparing the indicated group with the Asymptomatic group was <0.05.

The following predictors were found to be potentially associated univariately with the NVPQOL total or the SF36 summary scores: HG, multiparity, age, marital status, Hispanic ethnicity, education limited to high school, government health insurance, Beck Depression Inventory-II cognitive and somatic summary scores, history of motion sickness, unstimulated saliva volume, all three PUQE scores, the DAS summary score in the ‘cohesion’ domain and the independent individual Question 21 regarding quarreling, as well as the SCL depression, anxiety, somatic and global scores. Marin acculturation scores were noncontributory.

These potential predictors were entered into an RPA model for each of the HRQoL outcomes to identify the most at-risk HG and non-HG subjects (Appendix A). The RPA analysis was useful in characterizing the constellation of risk factors putting subjects most at risk for poor HRQoL. This was true when comparing the HG with the NVP groups. For example, 100% of women with HG who had a PUQE nausea score >2, and who were married or partnered, had a low total NVPQOL score, as did 100% of women in the NVP group who were highly symptomatic and multiparous. The RPA analysis was also helpful in understanding those who were most at risk within the HG or NVP groups. For example, 100% of women with HG who quarreled with their partner and who had a PUQE nausea score >2 had a low NVPQOL score in the Emotions domain, compared with 0% of women with HG who did not quarrel with their partner. The full RPA analysis for all domains included in both tests is shown in Appendix A, and illustrates those subgroups of patients who were most, and least, at risk for low HRQoL. Variables that surfaced in the RPA models associated with low NVPQOL scores were multiparity, Hispanic ethnicity, private health insurance and quarreling with a partner.

Figure 1 demonstrates an example of RPA analysis in the Fatigue domain of the NVPQOL, in which the subjects at highest risk for low quality of life were those with HG who had a high nausea score. Here, 94% of these women scored poorly in the Fatigue domain.

Figure 1.

Example of Recursive Partitioning Algorithm for Nausea and Vomiting of Pregnancy Quality of Life questionnaire (NVPQOL). ‘Fatigue’ domain. ‘LOW QUALITY OF LIFE’ represents an NVPQOL score <3 out of a possible seven for the Fatigue Domain. Pregnancy-Unique Quantification of Emesis (‘PUQE’) scores were split at a nausea score ≥3 out of a possible five points. These results are partially represented numerically in Table 3 for the Fatigue domain of the NVPQOL.

Table 3 describes the final results of the multivariate logistic regression models for each outcome, incorporating predictors and important interactions from linear regression models and RPA models.

Table 3.

Odds ratios for predictor variables for each of the health related quality of life outcomes as derived from the final logistic regression models.

Outcome	Variable	OR (95$ CI)	P-Value
LOW NVPQOL total (all domains) C=0.89	HG	2.91 (0.64–162.67)	0.1831
	Married/partnered	16.97 (1.77–162.67)	0.0141
	SCL-90 Somatic ≥60	22.71 (2.33–221.37)	0.0072
	PUQE Q1 (nausea) ≥3	68.37 (6.90–677.94)	0.0003
LOW NVPQOL physical domain C=0.82	HG	3.33 (1.04–10.68)	0.0428
	PUQUE Q2 (vomiting) ≥3	12.05 (3.39–42.81)	0.0001
	PUQUE Q3 (retching) ≥3	5.33 (1.50–18.99)	0.0098
LOW NVPQOL fatigue domain C=0.86	HG	6.06 (1.31–28.16)	0.0214
	Married/partnered	7.12 (1.41–36.07)	0.0178
	SCL Somatic ≥60	5.88 (1.21–28.66)	0.0283
	PUQE Q1 (nausea) ≥3	19.04 (3.75–96.75)	0.0004
LOW NVPQOL emotion domain C=0.80	HG	3.91 (1.08–14.17)	0.0382
	Age ≥3	8.98 (2.38–33.98)	0.0012
	History of motion sickness	4.99 (1.39–17.84)	0.0135
LOW NVPQOL limitations domain C=0.78	HG	3.26 (1.03–10.33)	0.0445
	PUQUE Q1 (nausea) ≥3	12.84 (3.50–47.11)	0.0001
LOW SF36 physical summary score C=0.77	HG	1.28 (0.30–5.47)	0.7377
	Hispanic ethnicity	0.19 (0.04–0.94)	0.0418
	SCL-90 Depression >60	5.13 (1.33–19.85)	0.0177
LOW SF36 mental summary score C=0.87	HG	2.66 (0.68–10.41)	0.1612
	PUQUE Q2 (vomiting) ≥3	4.90 (1.09–22.01)	0.0384
	History of motion sickness	3.81 (1.02–14.19)	0.0460
LOW SF36 physical function score C=0.53	HG	1.34 (0.41–4.47)	0.6284
LOW SF36 role physical score C=0.71	HG	4.93 (0.90–26.88)	0.0655
	History of motion sickness	4.06 (0.91–18.15)	0.0668
LOW SF36 bodily pain score (worse) C=0.75	HG	1.16 (0.29–4.69)	0.8344
	SCL90 Depression >60	8.04 (2.00–32.32)	0.0033
LOW SF36 general health score C=0.87	HG	2.53 (0.44–14.66)	0.3008
	Hispanic ethnicity	27.17 (1.39–529.51)	0.0293
	SCL90 Depression >60	21.66 (1.98–236.60)	0.0117
	Married/partnered	0.07 (0.01–0.77)	0.0306
LOW SF36 vitality score C=0.65	HG	0.87 (0.26–2.84)	0.8134
	PUQUE Q1 (nausea) ≥3	3.54 (1.09–11.53)	0.0360
LOW SF36 social functioning score C=0.79	HG	1.29 (0.27–6.05)	0.7506
	History of motion sickness	6.05 (1.18–30.97)	0.0222
	PUQUE Q2 (vomiting) ≥3	10.11 (1.39–73.46)	0.0307
	Private insurance (vs government)	9.25 (1.10–77.90)	0.0407
	No college education	14.61 (1.73–123.63)	0.0139
SF36 role emotional <40 C=0.60	HG	2.55 (0.61–10.63)	0.1985
SF36 mental health <40 C=0.55	HG	0.65, (0.21–2.01)	0.4503

Discussion

This exploratory study demonstrated two important findings. First, when testing with NVPQoL, women with HG seemed to have lower HRQoL scores when compared with women with NVP, and second, for both the NVPQoL and SF36 tests, perceived physical symptom severity and multiple psychosocial factors, such as depression, marital status and maternal age, seemed to be equally or more important contributors to low HRQoL than having an HG diagnosis.

This study represents a departure from previous studies in that it is the first to quantify the relative importance of the physical and psychosocial determinants of HRQoL for women with NVP, using both generic (SF36) and condition-specific (NVPQOL) HRQoL measures. Although Lacasse et al.⁸ demonstrated by using NVPQoL that the severity of NVP was associated with decreased HRQoL after controlling for multiple psychosocial variables, our study goes a step further, calculating an adjusted OR for HRQoL for women with HG compared with women with NVP using the NVPQOL and SF36 assessments. Furthermore, we include ORs for multiple psychosocial risk factors, after thoroughly exploring their interactions.

The hypothesis that women with HG (or severe NVP) would have a lower HRQoL was the same in both studies. However, although the study by Lacasse et al.⁸ supported this hypothesis, our study demonstrated that HG status did not fully explain the results. The association between HG and HRQoL was highly dependent on the test used (NVPQOL vs SF36) and the domain of interest (for example, Social Limitations vs Fatigue). Among women with NVP, the presence of some physical symptoms and psychosocial factors put them equally or more at risk than women with HG.

Overall, our results agreed with findings in the published literature that HRQoL, as measured by SF36, is low for pregnant women with NVP or HG in the first trimester compared with that for asymptomatic pregnant women and US population norms.

Physical symptoms

Given the published literature and the clinical impression that the more severe symptoms of HG should lead to decreased HRQoL, we expected to develop models for both NVPQOL and SF36, demonstrating that HG was associated with lower quality of life scores. For this study sample, the condition-specific instrument (NVPQOL) seemed to be able to demonstrate that women with HG had lower quality of life scores than did women with NVP, whereas the generic instrument (SF36) did not show any significant difference for HG in any domain. According to the NVPQOL instrument, women with HG were 3–6 times more likely to have a low HRQoL than women with NVP in each of the four domains, namely, Physical, Fatigue, Emotion and Limitations (Table 3). The RPA models for NVPQOL further supported the contribution of severe physical symptoms to poor quality of life, as they demonstrated that even women without an HG diagnosis who reported severe physical symptoms seemed to have low HRQoL scores. Severe physical symptoms are implicit for women with HG because these women have been diagnosed and hospitalized and their illness has been validated and treated. The NVP patients may perceive that they have serious symptoms, but without an HG diagnosis, the contribution of these symptoms to decreased HRQoL may be difficult to detect, unless these symptoms are measured independently. Our work broadens the understanding of the significance of what is considered ‘normal’ NVP, such that patients with NVP, similar to their HG counterparts, can experience an equally significant negative impact on HRQoL if they perceive their physical symptoms to be severe.

However, the diagnosis of HG itself was relatively noncontributory to the SF36 domains. HG was forced in the models, and, in general, ORs associated with HG were low. The highest OR for HG was 4.9 in the Role Physical domain (P=0.0655), and the other ORs ranged from 0.9 to 2.7 (P>0.15). Higher PUQE scores did contribute to lower quality of life in the Vitality, Social Function and Mental Summary SF36 domains, and did factor into some of the RPA models identifying women with and without HG who were most at risk for poor quality of life. These results confirm that the perceived severity of symptoms, and not necessarily the HG diagnosis itself, was associated with decreased HRQoL.

A history of motion sickness also seemed to contribute to some of the HRQoL models, and was associated with lower scores. This was evident in the NVPQOL Emotional domain (OR=5.0) and the SF36 Mental Summary score (OR=3.8), as well as in the Role Physical domain (OR=4.1) and the Social Functioning domain (OR=6.1).

Mental health symptoms

Many previous investigators have found an association between anxiety and depression and NVP and HG, although the causal direction of these relationships remains controversial.^13,37–39 Even normal pregnancy, especially in the early and final months, has been associated with these symptoms.⁴⁰ Although it has been suggested in literature, and accepted by many health-care professionals, that mental health symptoms were likely to precede the development of NVP and HG,⁴¹ more recent studies, such as that by Simpson et al.,⁴² provide a likely explanation that the psychological symptoms associated with HG may be due to the stress and trauma of illness; severe and debilitating physical symptoms are experienced by women affected with HG. However, any normal pregnant woman experiencing severe NVP, but not diagnosed with HG, can also be vulnerable to symptoms of stress, anxiety and depression. It is noteworthy in this study that only one patient had a previous history of treatment for depression, so that at least in this study sample, psychiatric illness did not seem to precede the onset of HG.

In this paper, mental health symptoms seemed to contribute to poor HRQoL scores. None of our available mental health measures seemed to be important in any of the preliminary or final models of the NVPQOL test. However, depression was a contributor to the SF36 physical summary score (OR=5.1), and to the Bodily Pain and General Health domains, with ORs of 8.0 and 21.7, respectively.

Social context

In keeping with Ware’s conceptual model for HRQoL, it has been demonstrated that lack of social support during pregnancy may exacerbate both mental and physical symptoms, and lead to a decreased HRQoL.^6,8,43 Married status has long been used as a measure of social support; however, the concept that marriage and marital stability are positively associated with health and well-being is now being challenged, and is now found to depend on the quality of the relationship and personality traits.⁴⁴ Furthermore, women’s adaptation to pregnancy in the presence of lower-quality relationships may also be buffered by the presence of social support.⁴⁵

In our study, multiple aspects of the social context within which NVP and HG occur seemed to be highly significant using either instrument. The final models for the NVPQOL assessment in Table 3 demonstrated that married or partnered women (in the QOL Total and Fatigue domains) and women over the age of 30 years (in the Emotional domain) were most at risk. Other variables that surfaced in the RPA models for the NVPQOL assessment that may have been indicative of social context, and perhaps of the level of social demand, were multiparity, Hispanic ethnicity, private health insurance and quarreling with their partner; these variables seemed to place women at risk for lower quality of life. It is interesting that the linear model for the total score on the NVPQoL test generated with an NVP population in the study by Lacasse et al.⁸ noted substantial parameters (adjusted β values) associated with similar variables that moved in the same direction, although they did not have enough power to be statistically significant. These variables were Hispanic ethnicity vs Caucasian race, living with spouse or with someone (for example, family or cotenant vs living alone) and having a parity of two or more vs nulliparity, all of which seemed to be associated with decreased HRQoL. Such potential relationships merit further exploration.

The SF36 final models (Table 3) incorporated social factors, although they did not always have the same effect as demonstrated in NVPQOL. Private health insurance placed women at risk for low scores in Social Functioning, consistent with the NVPQOL results. Hispanic ethnicity placed women at risk for low scores in the domain of General Health (OR=27.2), although paradoxically, such ethnicity did seem to be protective of the Physical Summary score (OR=0.2). Contrary to their at-risk status in the NVPQOL, married or partnered couples seemed to be protected in the General Health domain (OR=0.1). College education seemed to be protective under some circumstances (Social Functioning domain), and placed women at risk in others (Physical summary score).

The interpretation of the effect of social context seems quite complex, and will likely require extensive future investigation to achieve an understanding of the mechanisms by which social responsibilities and roles affect HRQoL in pregnancy. In a study of 124 participants with NVP, O’Brien et al.¹⁶ reported that almost any type of sensory perception could stimulate symptoms, which has implications for the substantial impact on employment, household and parenting responsibilities.⁶ The extent to which social context provides triggers that exacerbate NVP vs the extent to which NVP impacts social functioning remains unknown. Our results suggest that, at the least, women’s social context, including family obligations and mechanisms of financial support, cannot be ignored.

The clinical implications of this study are that some women with ‘normal’ NVP may be just as likely to have a low HRQoL score as women with HG. The health of women with NVP and HG would likely benefit from care by a multidisciplinary perinatal team, and these results underscore the importance of assessing HRQoL as a multilevel phenomenon, identifying and intervening in psychosocial and physical contexts. Addressing only physical symptoms, such as giving antiemetic medications, without offering emotional and psychosocial support, may be less than optimal care. Health professionals need to disseminate information about NVP so that family, friends and employers can provide the emotional and practical help necessary to lessen the burden of NVP.⁵

Implications for future research

This study has attempted to identify those women with NVP who are most at risk for poor HRQoL. It seems that the NVPQOL and SF36 assessments are measuring complementary aspects of HRQoL. Although both seem to measure various areas of social functioning, for this population, the NVPQOL seems to be more sensitive to physical symptoms, and the SF36 to psychiatric symptoms. Whether a single instrument and pertinent normative data should be developed for the identification and tracking of women at risk for poor HRQoL in pregnancy remains uncertain. Nevertheless, our results suggest many avenues for further exploration of who is most at risk for the development of poor HRQoL.

Strengths and limitations

This study attempted to quantify the degree to which women with HG may demonstrate lower HRQoL scores compared with those with NVP. We engaged in extensive exploratory modeling, which yielded OR values for the association of HG vs NVP, and multiple psychosocial factors, with the various dimensions of quality of life, and found that these other psychosocial factors were often more relevant. Nevertheless, our results may also be unique to our patient sample, and may be limited in the degree to which social factors were measured and interpreted. We had no measure of social support, employment, income or family responsibilities.

Women in the NVP and HG groups were recruited using different strategies, and although they were ostensibly from the same catchment population, unmeasured differences may have contributed to the complexity of the final models.

Furthermore, women who volunteered for the study who entered the NVP group may have had more severe symptoms than pregnant women in general, thus potentially minimizing the differences between the NVP and HG groups. Moreover, HG patients were recovered by the time they participated in the study and this may have also contributed to a potential underestimation of the difference between the HG and NVP groups.

Hyperemesis gravidarum is a rare condition, and although the study may lack statistical power to draw conclusions with respect to some analyses of interest, the final results, as displayed in Table 3, illustrate multiple potential drivers of HRQoL that seem to be very strong, with large OR in a variety of domains. Adjustment for multiple comparisons would affect confidence limits and P-values, but not the point estimates represented by these OR values. The models are complex and deserve exploration, particularly in the social context and in the meaning of social function and responsibility. Use of this information should inform the design of further studies conducted in various clinical settings.

Conclusion

Hyperemesis gravidarum and NVP not only affect the physical health of pregnant women, but can also negatively impact their psychosocial functioning. Identifying risk factors that mediate the known negative effect of HG on HRQoL can assist the multidisciplinary perinatal health-care team with early identification and management. Severe physical symptoms, in the presence or absence of an HG diagnosis, seem to be associated with a lower HRQoL. In addition to the contribution of physical symptoms to lower HRQoL, some women with ‘normal’ NVP seem to experience as significant a reduction in HRQoL as do their HG counterparts. Both physical and mental symptoms are likely affected by social roles and responsibilities, as well as by the quality of social support provided by the woman’s partner/spouse. Comprehensive biopsychosocial assessment and intervention by a perinatal health-care team are essential in addressing the HRQoL for women suffering from nausea and vomiting during pregnancy.