Development and validation of the Short Professional Quality of Life Scale based on versions IV and V of the Professional Quality of Life Scale

In some healthcare areas, professionals are vulnerable to the negative effects of the helping relationship such as compassion fatigue (CF) and burnout (BO), which affect their Professional Quality of Life. For example, Coetzee and Klopper [1] already noted in their work how a prolonged, continuous and intense contact with patient could lead compassion fatigue. In this same line, Dasan et al. [2] argued that organizational factors, such as resources, and individual factors, such as personality and coping strategies, could also favor the emergence of compassion fatigue. Prevalence rates of such phenomena range from 7.3% to 40% for high compassion fatigue in intensive care units [3], and around 60% for medium compassion fatigue in palliative care professionals [4]. In the case of burnout, the rates vary from 14.0% to 70.1% in the case of intensive care units [3], and around 30% of palliative care professionals have shown medium levels of burnout [4]. However, it is at times such as the COVID-19 health crisis, that the risk of suffering these negative processes tends to increase even more. The COVID-19 pandemic has increased pressure on health systems, forcing staff to make critical decisions in environments with multiple adverse conditions. Combined with the fact that their patients often have no therapeutic options, healthcare workers usually end up experiencing a consequent sense of failure. Over extended periods, this situation can lead health personnel to experience acute stress reactions from emotional overload, as well as other affective pathologies or psychosomatic responses. Thus, this health crisis is visibilizing a problem that is already occurring with different degrees of intensity and in different contexts. Understanding the impact that health emergencies have on the quality of care that professionals provide, as well as on their own well-being, requires the availability of valid and easy to apply cross-cultural comparison tools which allow the early identification of CF and BO and the calculation of its incidence. Thus, in the following research we aimed to test a short, 9-item version of the Professional Quality of Life (ProQOL) scale, based upon items from its versions IV and V.

As pointed out in the work by Stanton et al. [5], “the multivariate nature of modern organizational research, the apparent ‘survey fatigue’ of organizational members, and new demands to present survey materials online make shortened but psychometrically sound measures of organizationally relevant constructs worthwhile” (p. 168). Therefore, we developed a brief questionnaire based on the ProQOL scale which has been in use since 1995 and has been revised several times, with version V being the current one. In the present study, we used both versions IV and V of the ProQOL. There are hardly any differences between these versions, except for a few wording changes designed to improve their understanding. In Study 1, we used items from version IV of the ProQOL to examine the construct validity, measure invariance across three countries, and the reliability of this new short version. In Study 2, we used version V of the ProQOL to provide more evidence for the usefulness of the short version of this scale and to relate the new short ProQOL with traditional variables related to the quality of life of professionals.

BO is a syndrome that can be experienced by human services employees in stressful situations [6], with 60% reporting it during their careers [7]. Healthcare professionals are especially vulnerable to BO because their work context is characterized by high-risk decisions, dealing with the public, and expectations of compassion and sensitivity [8]. However, studies have shown that BO alone does not explain professionals’ emotional problems from working with individuals who are suffering or are in pain [9, 10]. In this context, CF, defined as the negative effects of working with traumatized people [11] has received increasing attention in recent years. CF focuses specifically on the chronic worry and tension produced by continued exposure to traumatized individuals [9]. Research on CF has also defined its opposite or converse effect, compassion satisfaction (CS). CS takes place when exposure to traumatic events produces gratification [12] from the joy that comes from helping others [10].

Specifically in healthcare professionals, lower levels of CF and BO have been related to a holistic practice of self-care [13], mindfulness [13], self-compassion [14], empathy [14], or coping with death [13], and CS has been positively related to these variables. For example, Sansó et al. [13] found significant, medium-sized relationships between coping with death and BO, CF, and CS (β = − 0.29, p < 0.01; β = − 0.28, p < 0.01; and β = 0.33, p < 0.01, respectively). In turn, Galiana et al. [14], found relationships between the dimensions of physical, psychosocial, and social self-care, and the three components of Professional Quality of Life. These were higher for the social self-care dimension, ranging from r = − 0.27 (p < 0.01) to r = 0.38 (p < 0.01), with lower values for psychological self-care, ranging from r = − 0.12 (p < 0.05) to r = − 0.37 (p < 0.01), and physical self-care, ranging from r = − 0.12 (p < 0.05) to r = − 0.22 (p < 0.01). Regarding self-compassion, Duarte et al. [15] reported correlations between self-compassion and CS of r = 0.32 (p < 0.01), of r = − 0.44 (p < 0.01) with BO, and r = − 0.30 (p < 0.01) with CS.

Various metrics have been developed to assess BO, CF, and/or CS. For instance, the review carried out by Bride et al. [11] examined six different CF scales to assess the different domains of CF. However, among the reviewed instruments, only the ProQOL scale assessed all the aforementioned constructs [16, 17]. The ProQOL is the most commonly used method for measuring the negative and positive effects of helping others who experience suffering and trauma. This tool assesses professionals’ quality of life via three subscales, the CS, CF, and BO, with each dimension being represented by 10 items. The ProQOL has been widely used across populations, with evidence gathered from informal caregivers [18], professionals working with people with intellectual disabilities [19], social workers [20], or health professionals [13, 21] in several countries, including Australia [21], Brazil [4], Canada [21], China [22], Israel [23], Spain [4], and the United States [24].

Despite its widespread use, evidence on the psychometric properties of the ProQOL is still scarce, with only five studies focusing on its validity. In 2015, Dang et al. [22] studied the reliability and validity of the ProQOL among Chinese governmental staff in areas stricken by the Wenchuan earthquake. They found reliability problems for the BO dimension and a poor fit for confirmatory factor analysis (CFA) model testing for construct validity. Similar results were found by Samson et al. [23] in a sample of Hebrew healthcare providers and found poor reliability results for the BO and CS subscales. However, because they used an exploratory approach, reliability estimates were calculated using a three-factor solution which did not comprise the original items. In 2017, Galiana et al. [4] studied the ProQOL structure in two samples of palliative care professionals from Spain and Brazil and were unable to find significant CFA factor loadings for some items on the BO and CF subscales in either sample, as well as poor reliability estimates for the BO dimension. Finally, and most recently, Hemsworth et al. [21] studied the validity and reliability of the ProQOL in three samples: Australian nurses, Canadian nurses, and Canadian palliative caregivers. They found reliability problems with the BO dimension and in the construct validity when they tested separate CFAs for each subscale in each sample. Problems with the model fit were found in the three models estimated for the BO subscale and in the residual indices of two of the three models estimated for the CF subscale. Finally, Heritage et al. [25] used Rasch analysis to examine the ProQOL in a sample of 1,615 Australian nurses and found a two-factor structure with the CS and CF dimensions, and advocated for a 21-item version which removed items from the BO and CF dimensions.

The availability of a tool that allows Professional Quality of Life to be monitored is of special importance among palliative care professionals such as physicians or nurses who work in environments with a high emotional demand requiring them to face death, loss, and grief on a daily basis. Of note, the quality of life of professionals not only affects the professionals themselves but is also required for quality of patient care. Such a tool should also meet certain criteria. First, the brevity of tools available to health managers and institutions to measure the quality of life of the professionals they oversee (in order to screen and detect potential problems, i.e., prevent BO and CF and enhancing their CS) is especially important. The conditions of working in healthcare contexts (and outcomes to be measured) mean that asking these professionals to answer long questionnaires further increases their workload and reduces their already valuable time. Indeed, as noted by Stanton et al. “longer surveys take more time to complete, tend to have more missing data, and have higher refusal rates than short surveys” [5] (p. 167). Second, the scale should be reliable and valid; we used the original long form of the ProQOL which has several psychometric problems, but only for some of its items. Third, the tool should provide measurement invariance data. As explained by Schmitt and Ali [26], scientists interested in measurement variance are concerned about the reliability and validity of measurement instruments used in different groups and situations. With increased globalization, the applicability of different measurement instruments in various cultures and multicultural and multilingual contexts is of interest both to companies and other institutions. As stated by Schmitt and Ali, “Differences (e.g., in culture, in language) in the populations being measured necessitate examining the degree to which the instrument measures the same construct across these groups” [26] (p. 327). This is to guarantee that individuals with the same construct level receive the same scores, regardless of their group membership. Considering the aforementioned, this study presents a brief measurement tool (the Short ProQOL) for assessing quality of life among professionals which is based on versions IV and V of the original ProQOL scale.

As Stamm et al. [43] recently expressed in relation to self-care for professionals in these times of pandemic, “Our work can be overwhelming. Our challenge is to maintain our resilience so that we can keep doing the work with care, energy, and compassion”. Therefore, straight-forward tools for monitoring Professional Quality of Life created by consensus approaches are more necessary than ever. The aim of this study was to present a new short version of the widely used ProQOL (versions IV and V). Thus, we tested the validity and reliability of the Short ProQOL among palliative care professionals from three different countries: Spain, Argentina, and Brazil in two different studies: in a total, more than 1,000 healthcare professionals.

Other short-version studies validated in samples of palliative care professionals and with good psychometric results are discussed in the scientific literature, for example, the Bugen Scale Of Coping With Death [40] or the Swedish Frommelt Attitudes Toward Care of the Dying Scale [44]. It is worth highlighting the importance of the availability of brief surveys for use in demanding work contexts such as end-of-life care in order to help protect these staff and not overload them with long questionnaires that further increase their burden and stress. Indeed, as already mentioned, an enormous range of conditions and outcomes can be measured in healthcare personnel but having to answer long questionnaires increases the workloads of these professionals and reduces their valuable time. Moreover, the use of long surveys is also associated with higher missing data and refusal rates [5].

To develop the new Short ProQOL, we combined information about the previous statistical performance of items from the long-version ProQOL, taking a theoretical approach. This is a common procedure for reducing long scales with problems with their psychometric properties. Previous studies focusing on the properties of the ProQOL reported problems with its reliability and validity, especially in the BO and CF dimensions [4, 21‐25]. Hence, we examined the 30 items on the ProQOL scale and retained those with no reported psychometric problems and which were consistent with Stamm’s [27] definition of BO, CF, and CS. Three items from each dimension were retained: items 10, 19, and 21 for BO; 9, 13, and 25 for CF; and 12, 18, and 30 for CS.

In Study 1, the new scale structure with the chosen items was tested in the three different samples (countries) using the short ProQOL version IV. The CFAs showed an adequate fit, thus offering evidence of the Short ProQOL construct validity. The structure tested maintained the three dimensions of quality of life among professionals (the BO, CF, and CS) proposed by Stamm [16, 17], but with only three items on each subscale. However, the retained items respected the original meaning of the dimensions, which focused on feelings of depletion and difficulties in dealing with work and a high workload (BO); being affected by others’ traumatic experiences and having symptoms similar to the ones observed in traumatized individuals, especially intrusive and frightening thoughts (CF); and satisfaction derived from work and from helping others (CS).

There was also evidence for the reliability of the Short ProQOL based on relevant estimates for the three dimensions or subscales. Compared to the previous results obtained with the long form of the ProQOL, this short version solved the reliability problems in the BO [4, 22] and CF dimensions [23] which had also been identified in previous shortened versions (i.e., [23]). In addition, the CFA model showed an appropriate factorial structure fit, therefore solving fit problems reported in the long version by Samson et al. [20] and Hemsworth et al. [21]. Finally, although the 21-item version presented by Heritage et al. [25] did not report any problems with reliability or validity, it is worth mentioning that these authors conducted their work in only one study of the scale (in an Australian sample). Here we provided evidence of the adequate behavior of the Short ProQOL in three different countries, speaking two different languages, and in two different studies (see the results of Study 2).

Once the structure was successfully tested in the three samples, our invariance analysis indicated scalar invariance of the Short ProQOL across the samples (countries). Although frequently understudied, measurement invariance is a core issue when making group comparisons when the groups can be understood as different populations (different countries, races, cultures, professions, etc.) [33]. To meaningfully compare relationships across groups, metric invariance is a necessary condition, while for group mean comparisons, scalar invariance is necessary [45]. However, these requirements were not met in most research regarding the ProQOL [24, 46‐59]. In our case, because the measurement invariance routine results were successful, we were able to test for mean differences. These results suggested that (1) BO was higher among the Argentinian professionals compared to the Spanish and Brazilian professionals; (2) CF was higher among Brazilians, compared to Spanish and Argentinian professionals; and (3) CS was the same among the three samples. Thus, these results suggest that this short survey using only nine items could be used in further research to capture differences attributable to context.

The differences in the age, sex, and distribution seen in the samples could have been because of different levels of BO and CF in relation to age, sex, or different professions of the participants. However, previous research has shown some controversy regarding differences in professionals’ quality of life in relation to these variables [24, 46‐59], and so future research should examine this possible explanation. Several studies have noted that sex is an important variable because women experience more BO than men [43, 44]. However, very few studies have examined sex differences related to CF. Of these, Van Hook and Rothenberg found that, in a sample of 175 child welfare workers (136 male and 28 female) with a variety of assignments, females reported more CF than males, with marginal significance [49]. Furthermore, an investigation conducted with ICU and oncology nurses, concluded that male nurses exhibited significantly higher CS and lower BO and CF than female nurses [50]. Similarly, one study assessing pediatric nurses in the USA [24] and another examining a sample of 1121 mental health professionals [51] concluded that female sex was associated with higher mean CS and CF scores than male sex. They also showed that psychiatrists experienced higher CF than other professionals.

However, another study [52] conducted among a total of 532 clinical social workers and psychologists, concluded that sex had no significant effect on CF. Moreover, results from the study conducted by Mooney et al. [50] showed that CF decreased with years of nursing experience, although this relationship did not fit with the other dimensions. A broader relationship was found in a cross-sectional study of registered nurses working in emergency departments throughout the United States [53]; they obtained significant differences in the Professional Quality of Life results for the CF, BO, and CS dimensions in relation to respondent age, but no statistical significance when comparing the difference between male and female nurses [53]. Additionally, the study conducted by Sprang et al. [54], examined 669 mental health workers and child welfare workers and found that males experienced significantly higher levels of CF compared to females.

Regarding differences among professions, most research indicates that the nursing profession is most affected by BO and CF. For instance, a recent study among professional care providers at Palliative Cancer Care Centers in India showed that nurses and nursing aids had significantly higher BO than the other professionals studied (physicians, social workers, physiotherapists, and pharmacists) [55]. This also fits with previous research indicating that the presence of BO and risk of CF is higher in nurses than in other health professions [56‐59]. Therefore, we concluded that the differences in CF and BO observed in our results may have been related to sex, because CF was highest in Brazilian professionals—the group which included the most women. However, the relationship of these dimensions with profession remains unclear, because BO was higher in Argentinian professionals—the group with the most physicians and least nurses, a result which contrasts with the literature we reviewed. Therefore, differences in BO across samples could be because of the effect of the country rather than the profession, and further research will be required to clarify this.

In Study 2, we used version V of the ProQOL for the new short version of the ProQOL; this version differed from version IV in five out of the nine items comprising the Short ProQOL version used in Study 1: the three BO dimension items and two of the CF dimension items. This second study confirmed that the short structure, with adaptations of these items, also showed evidence of an adequate factorial structure with adequate reliability indices and relationships with previous variables such as coping with death, self-compassion [15], and self-care. Competence in coping with death is key both for adequate professional development and Professional Quality of Life. There is also evidence to suggest that the absence of this capacity can involve emotional distress, BO, and CF, while its presence has been related to CS [4, 13, 60]. Moreover, recent research has found an association between higher levels of self-compassion and lower levels of BO and CF, as well as a positive relationship between self-compassion and CS. Indeed, Beaumont et al. [61] gathered evidence for an inverse relationship between self-compassion and BO in a sample of experienced psychotherapists and this same relationship was also found in a sample of nursing professionals [62].

Additionally, self-compassion promotes interpersonal skills and is related to other quality of life determinants among professionals, such as empathy [15, 63]. The proper practice of self-care was a key aspect for maintaining health and Professional Quality of Life [14]. Aukstinaityte and Zajanckauskaite-Staskeviciene [64] were some of the first authors to evaluate this, offering evidence of a negative relationship between self-care and CF which was also supported by evidence from work by Neville and Cole [65]. Moreover, Galiana et al. [14] and Sansó et al. [13] published evidence for the negative relationship between CF and BO, and the positive link between self-care and CS, both directly and indirectly. More recently, Sorenson et al. [66] reviewed published qualitative data and found that self-care was the most significant preventative measure that healthcare professionals had reported taking to protect themselves from developing CF.

Finally, we would like to mention that this study was limited by the relatively small sample sizes we used in the Brazilian and Argentinian cohorts, which could have affected the generalization of our results. Another limitation was that we did not study the content or face validity of the retained items. Although we examined the relationship of the items with Stamm’s original definitions for the dimensions, no specific analyses were conducted to test this. Regarding evidence of overlapping variance with the full form of the survey, we only gathered this data in one study rather than in independent administrations, as recommended in the literature [67]. Another shortcoming of this research was that there were significant differences in mean participant ages, and sex and profession distribution in each of the samples. Thus, these differences could also have been because BO and CF change in relation to age, sex, or different professions. Thus, evidence in this regard from future work would be welcomed.