Study of the measurement properties of the Portuguese Version of the Well-Being Questionnaire 12 ( W-BQ 12 ) in women with pregnancy loss

The assessment of well-being can be an important parameter in monitoring the process of mourning. In this study we sought to assess the applicability of the W-BQ12 – Well-Being Questionnaire in a sample consisting of 74 women between four and six weeks following their respective pregnancy losses, analyzing its measurement capabilities. As proposed by the author, the analysis of the scale’s primary components and the discriminating convergent validity confirmed the dimensional structure of three subscales. The scale showed good reliability (global Cronbach’s Alpha coefficient = 0.84), and, overall, the items showed a good correlation with the corresponding subscale. Generally speaking, the W-BQ12 showed good discriminative validity when correlated with the PBGS Perinatal Bereavement Grief Scale. This tool, applied to this study, was found to be both reliable and valid for use by nurses in assessing the well-being of women who have experienced this type of loss.


Introduction
For a woman, involuntary pregnancy loss, besides involving psychological stress, can be an important factor in the development of a wide range of physical and emotional upheavals.This situation, as a negative event, deserves greater attention due to its impact, both potential and negative, on well-being.Rarely is it clear exactly what has been lost, given that the possible loss of self-esteem, loss of status (as a woman and potential mother) and existential loss is inherent to the loss of a pregnancy and a loved one, whether real or imaginary (1) .
The bereavement process is complex and varies widely from individual to individual in its duration.However, desirable reactions are those which, although painful, guide the woman in the sense of accepting the loss and finding the strength to continue her life (2) with self-confidence and a feeling of security, contributing to positive affect and well-being, which can constitute an important parameter in the monitoring of resolution of mourning.
The literature recommends the use of reliable instruments, whose results may be compared internationally.Apart from their transcultural adaptation, it is crucial for these instruments' measurement properties to be validated in the different populations which will use them.With a view to achieving this, the authors intend in this study to evaluate the measurement properties of an instrument for measuring well-being, which may constitute an important contribution to the evaluation of the process of mourning among these women.

Objective
To validate the measurement properties of the Portuguese version of the Well-Being Questionnaire12 (W-BQ12), in a sample of women who have experienced involuntary loss of a pregnancy.

Method
The W-BQ12 was validated by means of a set of procedures so as to evaluate its metric equivalence.
This was undertaken by the analysis of its reliability and construct validity for the Portuguese population.
A scale's measurement validity is the guarantee that the test measures what it claims to measure (2) .The scale's validity as a construct was inspected by means of the factorial organization of its items, using Principal Component Analysis, following the procedures adopted by the authors.
So as to confirm the organization of the scale in the study's sample, the authors analyzed the items' convergent and discriminant validity, by means of Pearson correlation between the items and the different subscales, with the item-subscale discrimination criteria defined as a correlation value greater than 0.10.The convergent validity "...refers to the extent to which the correlation of the instrument with instruments which measure the same construct is greater than with those which measure different constructs" (2) .It was in this context that the authors analyzed the Pearson correlation values between the scale under analysis and the Perinatal Bereavement Grief Scale (PBGS), filled out at the same time, so as to perceive the relationship between the theoretical constructs which they evaluated.
For the analysis of the scale's reliability, the internal consistency of each of the subscales and of the global scale was analyzed by the authors, considering Cronbach alpha values higher than 0.80 to indicate good internal consistency (3) , although values above 0.60 would be acceptable in scales with a reduced number of items (2) .

Instruments
The W-BQ was originally conceived, in 1982 (4)(5) , as an instrument for measuring states of depression, anxiety and some aspects of positive well-being, and was applied in a World Health Organization (W.H.O) study for evaluating new treatments for the management of diabetes.Although the initial version of 22 items continues to be recommended for studies in these areas, later, as a result of studies developed by the author and her collaborators (4)(5)(6)(7) a reduced version of 12 items was brought out, derived from the initial scale but designed for measuring psychological wellbeing.More recently, a longer version has been produced, with 28 items, incorporating a subscale for stress which is specifically for individuals with diabetes (ibidem).
Although this instrument, in its different versions, was designed for studies on diabetics, its applicability is not restricted to this area.It has been translated into more than 35 languages (5) , having been used for various kinds of studies in different cultures (5)(6)(7)(8) .The Portuguese version was made available to the researchers by the author*, upon the establishment of a protocol.
The W-BQ12 takes the form of 12 statements, which people grade according to the frequency of their occurrence over a period understood as being between the fourth and sixth weeks after the loss of the pregnancy, using a Likert-type scale of four points, in which 0 corresponds to never and the value 3, to always.It is composed of three subscales of four items each: one is intended to evaluate negative well-being (items 1 to 4), another is energy (items 5 to 8) and, lastly, one which evaluates positive well-being (items 9 to 12).The energy subscale is the only to present a mixture of items of a positive nature (items 5 and 8) and negative (items 6 and 7), implying a respective inversion of the two in the statistical procedures.Each subscale's score can vary between 0 and 12.The score for the total wellbeing scale is the sum of the 12 items (after inverting the scores for items 6 and 7) and can vary between 0 and 36.
High scores indicate a high perception of well-being.
To evaluate the convergent validity of the W-BQ12, the authors used the Perinatal Bereavement Grief Scale (PBGS), developed for studies on perinatal mourning (9) , translated, adapted and validated for the Portuguese population (10) , and designed to evaluate mourning and anxiety, or feelings of missing the pregnancy or the baby which was lost.The scale is in the form of 15 statements and the idea is that participants respond about the frequency (not the intensity) with which they have occurred in the last week, using a Likert-type scale of four points, in which the value 1 corresponds to rarely or never (less than once a day), the value 2 corresponds to sometimes (one or two days), the value 3 to many times more intense manifestation of mourning, or, as it may be, mourning which has resolved less.

Participants
For For 91.9% (n=68) of the interviewees this had been an accepted pregnancy, while for the others (8.1%, n=6) the pregnancy had either been not accepted in the beginning or never had been.Of the total of the participants, 64.9% (n=48) had experienced a spontaneous process of expulsion of the product of conception, while 35.1% (n=26) needed some type of surgical intervention.

Procedures
The necessary contacts with the author were

Results
The structural dimensions, the validity and the reliability were assessed, with the aim of evaluating the instrument's metric equivalence in the populational sample of the present study.

Validity of the Construct
With a view to understanding how the contents of the items in the scale are conceptually organized, the researchers proceeded to an exploratory analysis of the principal components, obtaining four factors.Later, and with a view to bringing the scale closer to the original version, this was reduced to three, with only items with a factorial load equal to or superior to 0.30 being selected for analysis, in accordance with the literature (11) .
The Varimax rotation method with Kaiser normalization was chosen so as to maximize the saturation of items.The factorial solution thus obtained explains 61.84% of the scale's total variance, with factor 1 (Negative well-being) best explaining the scale's variance (36.17%).The distribution of the items' factorial load can be observed in Table 1.

Convergent-Discriminant Validity of the Items
With the intention of confirming the structure of the scale of the sample under study, the authors proceeded to the analysis of the items' convergent-discriminant validity, using the Pearson correlation between the items and the different subscales.The values obtained for each item are shown in Table 2.
Having as a criteria that the correlations of the items with the subscale to which they belong should be superior to 0.10 relative to their correlation with the remaining subscales and, analyzing the items which reveal themselves to be problematic in the analysis of principal components, it can be said that items 5, 7 and 8 do not present discriminative power among the subscales.

Item 11 (I feel enthusiastic to get on with daily tasks
or take new decisions), which was also problematic in the principal components analysis, maintains a greater correlation with the energy subscale, with a difference greater than 0.10 relative to the subscale to which it originally belonged -positive well-being.In this respect, and given that convergent-discriminant validity analysis of the items reinforces the results obtained by the principal components analysis, the authors considered the possibility of item 11 moving from the positive well-being subscale to the energy subscale.However, the authors also tried the possibility of associating the constructs in a more appropriate way, in two subscales, such that the principal components analysis was conducted again, reducing to two factors.The resulting factorial solution was not satisfactory, given that the items considered the most problematic remained as they were, and finally the authors decided to keep the three-dimensional structure of the W-BQ12, altering item 11 from the positive wellbeing subscale to the energy subscale.

Convergent Validity
So that it might be possible to evaluate the convergent validity of the constructs under evaluation in the different subscales of the W-BQ12, the authors calculated its correlation with the PBGS filled out simultaneously.
Observation of Table 3 permits one to verify a negative correlation between the perception of general well-being and mourning, that is, the greater the perception of general well-being, the lesser the expression of mourning, that is, mourning is more resolved, corresponding to what is expected.Relative to the perception of negative well-being, one may observe a positive correlation with mourning.The correlation of perception of energy and positive well-being with mourning is not statistically significant.
As a result of the application of the instrument under analysis, and with the aim of understanding the perception   (5,8) .
It was concluded that the least robust subscale was that of energy, due to the conceptual structure of its items, which include characteristics which, in Portuguese culture, may be reflected in energy levels but also in well-being.
The difficulty in discriminating between some items of the energy subscale with positive well-being and negative well-being may be due to the fact that the same subscale presents one dimension of positive energy and another of negative energy and that the respondents, in the situation of the study, had some difficulty in disassociating the two concepts.
It is the negative well-being subscale which presents the best internal consistency; in contrast, the positive wellbeing subscale presents a weaker internal consistency (3) , given that its value is between 0.60 and 0.70 in spite of being very close to the higher value, although values over 0.60 may be acceptable (2) and justifiable, above all if the scales have a reduced number of items.Despite the values verified in each subscale individually, the general well-being shows strong internal consistency.
The W-BQ12 presented consistent results when correlated with PBGS with the existence of two different, although inter-related, constructs (subjective well-being and mourning).A negative correlation was observed between general well-being and mourning; that is, the greater the perception of well-being, the lesser the expression of mourning, or, as it were, the more resolved mourning was.On the other hand, the negative wellbeing subscale presented a positive correlation with mourning; that is, it is associated with a stronger, less resolved sense of mourning.
The loss of a pregnancy, with all its implications, as a stressful event, can interfere with women's perception of their well-being.As the degree of the sadness diminishes, and the mourning is worked through, the level of wellbeing gradually becomes more positive.
Relative to the subscales energy and positive wellbeing, one may observe negative correlations with the degree of mourning, but these are not statistically significant.
The analysis of the present study's results permits one to conclude that the participants presented satisfactory levels of positive well-being, general well-being, and energy, compatible with a process of mourning which is in resolution, revealing satisfaction with their personal lives, and enthusiasm and confidence that they can continue along life's journey.

Conclusion
According to the results obtained, one can say that the Portuguese version of W-BQ12 proved itself a reliable and valid measure for evaluating well-being in women who have involuntarily lost a pregnancy, given the reasonable values for internal consistency of the subscales and of the global scale, as well as the structure resulting from the factorial analysis.It was shown that the questionnaire was sensitive for a population different from that in which it had previously been used, which is an indicator of its validity.Nevertheless it would be interesting to consider its applicability in a larger sample.
The use of such an instrument permits one to evaluate and identify with greater precision the women at risk of a prolonged or more complicated mourning process, after the involuntary loss of a pregnancy.
3 to 4 days) and the value 4 to all the time or almost all the time (5 to 7 days).The PBGS score is a simple arithmetical sum of the individual scores of the items, and can vary between 15 and 60.A high score represents a *Clare Bradley.W-BQ12 © Portuguese version for Portugal.2005.Health Psychology Research, Dept of Psychology, Royal Holloway, University of London, Surrey.
established and a protocol was signed for use of the Portuguese version of the W-BQ12.The present study took place in public hospitals in the north of Portugal after obtaining approval from the respective ethics committees and authorizations from service management, as well as from the Executive Boards.The possible participants were contacted personally by the principal investigator, during the short period of hospitalization (average length, 24 hours) resulting from the loss of the pregnancy.This involved informing them about the objectives of the study and inviting them to participate, following all the informed consent procedures, in accordance with the Helsinki Declaration.The second contact was made between four and six weeks after the loss of the pregnancy, by means of a telephone interview, such that the researchers were able to evaluate the woman's perception of her well-being at that moment by applying the Portuguese version of the W-BQ12.The PASW program Statistics 18 was used for statistical processing of the data.

Table 1 -
Analysis of the W-BQ12's Principal Components,

Table 2 -
Convergent-discriminant validity of the items in Note: Values of r at the level of p <0.05* and p <0.01 † for N = 74 are considered significant; the values of the correlation relative to each subscale were corrected for overlapping

Table 3
Note: Values of r at the level of p < 0.01 (*) are considered significant