The quality of close relationships and the sense of social connectedness are reliable predictors of health and longevity (Holt-Lunstad et al., 2017; Schetter, 2017). Not surprisingly, the WHO enlisted social support networks among the most critical determinants of health (Commission on Social Determinants of Health, 2008). Social support networks perform four main functions: Emotional, Informational, Appraisal, and Instrumental. Briefly, Emotional support (also called affective or attachment) refers to demonstrations of love and caring, encouragement, and empathy from which one derives a sense of security (Thoits, 2011). Informational support is defined as providing facts or advice that may help a person solve problems (Wang et al., 2014). Instrumental support is intended as a form of practical, tangible help, provided through material assistance or practical tasks (Langford et al., 1997; Wang et al., 2014). Last, Appraisal support consists of expressions that affirm the appropriateness of acts or statements made by another (Langford et al., 1997).

Both informal and formal networks can support people affected by chronic diseases. Thus, this article refers to social support activated or required to cope with adverse life events and chronic health problems, defined as problem-oriented social support (e.g., Suurmeijer et al., 1995). Research has consistently demonstrated several benefits that medical patients receive from their informal networks (e.g., family, friends, and relatives). In conditions like diabetes, these benefits include improving emotion regulation, coping, glycemic control, and quality of life (Van Dam et al., 2005; Strom and Egede, 2012; Hill-Briggs et al., 2021). Social support also enhanced functional status and quality of life in people with heart failure, influencing treatment and health management behaviors (Graven and Grant, 2013, 2014; Bucholz et al., 2014).

In oncology patients, social support was found to address psychological problems resulting from a poor adjustment to cancer and its treatment (Rizalar et al., 2014). For example, by fostering acceptance, positive reframing of the situation, and maintaining the patient’s sense of humor, social support networks help patients be more determined in their fight against the disease and counteract helplessness-hopelessness and anxious preoccupations (Kawa, 2017; Lauriola and Tomai, 2019; Tomai et al., 2019). Moreover, several authors (Pinquart and Duberstein, 2010; Ikeda et al., 2013; Yağmur and Duman, 2016) reported significant correlations of social support with “hard” health outcomes, such as tumor initiation, cell proliferation, and life extension.

It is worth noting that social support may unintentionally hurt the patient’s wellbeing. Previous research has indicated that, contrary to caregivers’ intentions, patients who report unmet needs may perceive some forms of support as ineffective or troublesome (Breuer et al., 2017; Sebri et al., 2021). In particular, social support turns out unsupportive when the helper does not understand or effectively correspond to the recipient’s desires (Nouman and Zanbar, 2020). For example, this can happen within close relationships, when help is not sought or when support becomes controlling, oppressive, or, conversely, too superficial and neglecting the consequences of illness (Mazzoni and Cicognani, 2016; Mazzoni et al., 2017). Therefore, paying attention to the type of support a patient needs and how that support is provided will be the best way to address patients’ specific needs.

In addition to family and friends, healthcare professionals (e.g., doctors and nurses) can provide social support to patients suffering from chronic diseases, thus becoming a formal social support network. For instance, a supportive healthcare provider helped diabetic patients to defuse health distress and improve glycemia (Venkatesh and Weatherspoon, 2013; Wardian and Sun, 2014). Similarly, according to a recent study (Ban et al., 2021), the medical staff reduced cancer patients’ fear of illness progression by providing social support. Indeed, identifying patients at risk for psychosocial vulnerability due to low healthcare support appears to be a relevant health outcome (Usta, 2012).

How do formal social support networks work in oncology units? Are there similar functions that informal and formal networks perform for cancer patients? Healthcare social support functions have been less extensively studied than informal network ones. Nevertheless, given the increasing interest in improving healthcare quality and patient experience, constructs similar to social support functions have been used in medical settings (e.g., Donabedian, 2005; Beattie et al., 2015; Hancock et al., 2020). For instance, healthcare quality indicators include ratings of physicians’ interpersonal skills and information provision (Brédart et al., 2005a,b). Trust in oncologists reflects the patients’ beliefs about the healthcare provider’s ability to provide appropriate, reliable, and hopefully successful treatment (Müller et al., 2014). Communication is also considered a core clinical skill for establishing an excellent patient–doctor relationship (Malley and Fernández, 2010). When clinicians are better communicators, they can effectively convey information to patients, encouragement, and tangible support (Street et al., 2009). In sum, a health professional’s interpersonal competence and patient trust in the healthcare provider might be associated with one’s perception of social support. Still, they are not part of the construct as defined in psychosocial research (Barrera and Ainlay, 1983).

Where scholars have explicitly addressed healthcare social support, methodological approaches to obtaining reliable and valid measures have been varied and scattered. Previous research lacked a clear connection with the fourfold structure of the construct. For instance, some studies assessed multiple aspects of social support but used a single global score (Katz et al., 2003; Reynolds and Perrin, 2004). This choice, however, does not make fine-grained distinctions between how the healthcare provider can support the patient and improve the therapeutic relationship. Other studies focused either on emotional support (Kuuppelomäki, 2003; Wenrich et al., 2003; Ansmann et al., 2012) or informational support (Rutten et al., 2005). Underlying this approach, emotional and informational supports are viewed as functionally independent and not interchangeable. Another original approach combined two questions for each type of social support function into a single dichotomous indicator (Arora and Gustafson, 2009), assuming that healthcare provider social support could be a categorical variable. Unclear boundaries exist between informal networks’ social support functions and other proxy constructs used in medical research (e.g., communication skills).

Healthcare social support depends not only on individual actors (e.g., doctors or nurses) but also on the hospital’s mission and organizational culture. Ansmann et al. (2014), for example, emphasized the importance of hospital characteristics in fostering a better patient-doctor relationship. Hospitals in which the local culture emphasized cost control typically reported greater dissatisfaction with physicians’ interpersonal skills and information provision (Zhou et al., 2011). By contrast, non-profit hospitals imbibed in a religious culture promote a more inclusive and respectful atmosphere (Whitley, 2012). Religious hospitals have traditionally been reputed to provide a higher quality of care than government-operated hospitals (Fleming, 1981). Indeed, the organization’s charitable mission to serve and care for the person, along with a religious institutional identity, promotes the humanization of health care (e.g., Reinikka and Svensson, 2010; Calegari et al., 2015). Thus, religious hospitals can offer close and sensitive listening to the patient’s narratives and more compassionate care services, including care for the poor and vulnerable (White and Begun, 1998; White et al., 2010). Users themselves describe religious hospitals as more reliable and attractive than government-operated ones (Seemann et al., 2015). Considering this literature, a healthcare provider social support measure could also be a valuable tool to compare hospitals, outpatient clinics, particular services, or departments.

Investigating the factor structure of a new scale is vital to establishing its psychometric properties. Confirmatory Factor Analysis (CFA) is commonly used for this purpose. Nevertheless, CFA has recently complained about failing to account for the imperfect nature of items, as evidenced by significant item correlations with non-target constructs or cross-loadings on more than a factor (Marsh et al., 2014; Morin et al., 2016). As a result, CFA may produce inaccurate estimates of factor correlations, necessitating general or method factors to achieve an acceptable fit (Morin et al., 2016; Joshanloo et al., 2017). In the present study, we used Exploratory Structural Equation Modeling (ESEM), a new approach that overcomes the limitations of CFA (Asparouhov and Muthén, 2009). In ESEM, cross-loadings on non-target factors are allowed in addition to primary loadings on target factors. Thus, ESEM is more efficient than CFA in the estimation of complex models because it provides more accurate estimates of factor loadings and correlations, preventing artificial inflation of factor loadings on the general factor (Marsh et al., 2014; Morin et al., 2016). We further examined the HPSS dimensionality and reliability using Mokken Scaling Analysis (MSA) to provide a sensitivity analysis for ESEM results. As shown in Figure 1, we tested the following factor models, each based on different theoretical assumptions. Like previous studies (Katz et al., 2003; Reynolds and Perrin, 2004), the unifactorial model assumes that all HPSS items measure the generalized patient’s perception of the healthcare provider’s support (Figure 1A). The four-factor model is consistent with the view that patients could discriminate between ways a doctor had supported them (Figure 1B). Assuming correlated factors would imply that the four social support functions were somewhat interrelated. Small factor correlations would show that different types of support are distinct but related aspects. Moderate factor correlations would indicate substantial overlap, suggesting a hierarchical arrangement of factors. In a bifactor model, every item is targeted to load primarily on general and specific factors (Figure 1C). For example, an item describing an emotionally supportive act is hypothesized to load on a generalized support perception factor and a specific emotional support one.

As a by-product of the study, we explored whether the HPSS ratings could detect differences between two healthcare providers. In keeping with the literature (Zhou et al., 2011; Whitley, 2012; Ansmann et al., 2014), we expected that patient ratings of perceived social support on the HPSS would differ between two oncology centers reputed for religious and government-operated organizational culture, respectively. Last, we explored whether the HPSS was sensitive to how patients at different stages of disease perceived their healthcare provider as providing support for all types or just for specific kinds. This hypothesis was motivated by studies showing that a patient’s need for support from a healthcare provider may change during the cancer experience. For instance, late in the course of cancer, especially when patients perceive a poorer prognosis, there is a greater need for emotional and instrumental support from health professionals (e.g., Arora and Gustafson, 2009).

The previous study showed that the HPSS scale has a solid factorial structure and that the four social support functions can be reliably measured. However, examining only the factorial structure of the scale is not sufficient to demonstrate the validity of HPSS scores. Given that validation of a new measure is a laborious process that requires a range of empirical evidence, in this study, we surveyed an independent sample of oncology patients to take the first step in this direction. As part of criterion-related validity assessment, we aimed to establish the correlations between HPSS scores and two critical variables in the physician-patient relationship. These were the physician’s communication skills and the patient’s trust toward the healthcare provider. Recent studies have reinforced the view that patients of health professionals with better communication skills develop trusting relationships with their doctors (Chandra et al., 2018) and have better health outcomes (Howick et al., 2018; Noble, 2020). Accordingly, we expected the HPSS scores to positively correlate with the doctor’s communication skills and the patient’s trust toward the health care provider.

In the present study, we proposed a new scale, the HPSS, to assess healthcare social support in oncology settings. The scale was designed according to a multidimensional approach, as established in psychosocial research where social networks perform Emotional, Informational, Appraisal, and Instrumental functions (Cutrona and Russell, 1987; Langford et al., 1997; Thoits, 2011; Wang et al., 2014).

Although still preliminary, the results obtained from two studies of patients recruited from different oncology day-treatment units are promising and can be summarized as follows. First, the HPSS had a multidimensional structure. Second, the HPSS subscales proved reliable and preserved a good degree of specific information. Third, we detected differences in perceived social support between patients admitted to religious and government-operated hospitals. Fourth, the Affective, Informational, and Appraisal functions were positively correlated with doctor communication skills and patient’s trust in the physician.

Regarding the factor structure, our study showed that the four-factor model had a good fit to the data, outperforming the unifactorial model. Other studies measured healthcare provider social support either as Emotional or as Informational, implicitly assuming that these functions were enough to cover the content domain and precluding comprehensive tests of construct dimensionality (Kuuppelomäki, 2003; Wenrich et al., 2003; Rutten et al., 2005; Ansmann et al., 2012). Our findings indicated that oncology patients could discriminate how a healthcare provider could support them beyond merely demonstrating love and caring, encouragement, and empathy (Thoits, 2011) or providing facts or advice (Wang et al., 2014). Instrumental and Appraisal emerged as distinct social support factors. In a doctor-patient relationship, it could be argued that providing practical, tangible help to patients (Langford et al., 1997; Wang et al., 2014) or endorsing the appropriateness of acts or statements made by patients (Langford et al., 1997) is not required. Nevertheless, even assuming this is true for all medical specialties, it is still possible that other health professionals (e.g., nurses or physical therapists) might contribute to enhancing the perceived quality of care by providing complementary functions of social support. Foreshadowing future research, one could investigate whether Instrumental and Appraisal items apply equally well to physicians and other members of a multidisciplinary oncology care team.

Our study also revealed that a general factor coexisted with the abovementioned factors. CFA has recently complained about necessitating insubstantial or inflated general factors to achieve an acceptable fit (Morin et al., 2016; Joshanloo et al., 2017). Because we used an ESEM approach, which does not produce such statistical artifacts, we are reasonably confident that the general factor represented an additional source of reliable variance. However, the general factor could be challenging to explain. Alternative interpretations are possible. On the one hand, the general factor might reflect a “true” perception that accounts for patients’ overall feelings of being supported by healthcare professionals. This interpretation is consistent with previous research considering healthcare social support as a single evaluative dimension, disregarding fine-grained distinctions between specific functions (Katz et al., 2003; Reynolds and Perrin, 2004). On the other hand, the general factor could reflect acquiescence, social desirability, or other response sets. We believe that the first interpretation is the most likely. However, we cannot rule out the second interpretation based on the present study. Thus, future research should validate the general factor against independent response-bias measures.

Although our study supported the construct validity of the HPSS, a few items showed low loadings on the target factors or lacked factorial simplicity in ESEM analysis; a few others were shown to violate Mokken scaling assumptions. As noted elsewhere (e.g., Morin et al., 2015), the inherent difficulty in producing items that perfectly reflect the constructs intended to measure could explain these defects. However, we cannot exclude other possibilities. For example, in developing the HPSS, we adopted a mostly etic approach to content validity and item generation. However, the items also reflected an emic view of patients in the illustrative examples provided below each item (see Supplementary Materials). While combining etic/emic approaches is desirable to widen the content coverage (Magasi et al., 2012), the specific item format (i.e., general statement + illustrative examples) might have increased the item complexity. Of course, defective items might be dropped out from a revision of the HPSS; however, we did not find severely biased or insubstantial items in the present study. So, we think modifying items could be more appropriate than a deletion. Notably, notwithstanding imperfections, the subscale scores preserved a non-negligible amount of reliable information and were reasonably scalable. Therefore, we recommend using subscales in clinical assessment and research applications.

The present study also made some steps toward a more robust assessment of criterion-related and concurrent validity of the HPSS scores. In keeping with previous research (White and Begun, 1998; Reinikka and Svensson, 2010; White et al., 2010; Zhou et al., 2011; Whitley, 2012; Calegari et al., 2015; Seemann et al., 2015), we expected the HPSS to be sensitive to differences in perceived support provided to patients admitted to religious or government-operated hospitals. This hypothesis was supported for Emotional, Informational, and Instrumental support, perceived higher by patients in the religious hospital. This finding implies that the HPSS could be used in quality of care research to assess and compare hospitals, outpatient clinics, particular services, or departments. As part of criterion-related validity, we also expected stage-related differences in perceived social support as suggested by previous research (Arora and Gustafson, 2009). This hypothesis was not confirmed, however. Either the patients have not changed their demands on the doctors, or the doctors may not have adapted to the patient’s changing needs. A longitudinal study would be needed to address this issue.

Construct validation is a long-term endeavor, requiring multiple studies and accumulating evidence for the instrument’s validity. Our study aimed to establish the correlations between HPSS scores and two critical variables in the physician-patient relationship as part of this process. Constructs similar to social support functions have been used in medical research (Donabedian, 2005; Brédart et al., 2005a,b; Malley and Fernández, 2010; Müller et al., 2014). From our perspective, a physician’s communication skills and trustworthiness are not social support variables; nevertheless, we expected to see positive correlations between HPSS scores and these variables. Overall, our hypotheses about the relationships of HPSS ratings with similar constructs in the nomological network were confirmed.

Beyond merely providing initial evidence of HPSS concurrent validity, the correlation pattern reported in the present study was compatible with the view that healthcare provider social support was needed to develop a trusting doctor-patient relationship, above and beyond the physician’s excellent communication abilities. Indeed, Affective and Informational scores remained statistically associated with patients’ trust in their physician, controlling for the physician’s communication skills. Not only were the patients of doctors with excellent communication skills more likely to develop trustworthy relationships (Chandra et al., 2018), but specific healthcare support functions can also have a unique role in increasing patients’ trust.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Studies involving human participants were reviewed and approved by the Hospital R Ethics Committee (Study 1) and the Ethics Review Committee for Psychological Research, Department of Social and Developmental Psychology, Sapienza University of Rome (Study 2). The Hospital R ethics committee waived the requirement for written informed consent for participation in Study 1. We obtained written informed consent from all patients participating in Study 2 as per the Ethics Review Committee for Psychological Research recommendations.

MT and ML contributed equally to the theoretical and empirical aspects of the study. All authors contributed to the article and approved the submitted version.

This work was supported by the Sapienza Institutional Research grants (RP1181642870C00B, RM11916B7E07E816).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.