Effectiveness, structure, and content of nurse counseling in gynecologic oncology: a systematic review

A comprehensive literature search [28, 29] was performed in PubMed®, CINAHL®, PsychINFO®, Cochrane®, and EMBASE® until July 2016. The search strategy (Fig. 1) was developed for PubMed and then adapted for all other databases. To identify additional studies, reference lists of relevant articles were hand-searched and authors contacted.

Inclusion criteria for retrieved records were English publications of experimental (randomized controlled trials), quasi-experimental (pretest-posttest studies and time series with non-equivalent comparison groups), or pre-experimental (one-group pretest-posttest studies and posttest-only studies with comparison groups) studies assessing the effectiveness of nurse counseling. Nurse counseling was defined as a problem-solving approach to support patients by a voluntary, interaction-intensive orientation, planning, and decision process. All outcomes of women with gynecological pre-cancer confirmed by colposcopy/biopsy (cervical, vulvar, or vaginal intraepithelial neoplasia) or with gynecological cancer (of the cervix, vulva, vagina, endometrium, ovary, or fallopian tube) were eligible. Interventions not mainly provided by a nurse and those designed for palliative patients, children, or adolescents were excluded.

After removing duplicates, two authors (SR and AK) independently screened the titles and abstracts of all records and, subsequently, the full-text versions of all remaining records for eligibility. They documented the inclusion and exclusion criteria for each study and discussed disagreements until consensus was reached.

Subsequently, the two authors independently assessed the methodological quality of included studies, whereby a statistician supported the appraisal of appropriate statistical analysis. Experimental studies were assessed with the Joanna Briggs Institute 2014 [30] (JBI) tool for Randomized Control/Pseudo-randomized Trials and for all other studies, the JBI tool for Descriptive/Case Series Studies was appropriate (Additional file 1). The sum quality score for each study was converted into percentage values and interpreted in accordance with studies based on similar methodologies (e.g., [31, 32]). Values from 0 to 49% were defined as of low methodological quality, and those from 50 to 79% and from 80 to 100% as of moderate and high methodological quality, respectively. The two authors compared their assessment results against one another and discussed any disagreement until consensus was reached. Finally the evidence level (Additional file 2) of each included study design was determined with the JBI classification [33] (from level 1, experimental designs, to level 4, observational-descriptive designs).

After this assessment process, the first author extracted the characteristics and results of all included studies and contacted the study authors to gather any missing information in this regard.

For data synthesis, the effect of nurse counseling on each identified patient outcome was narratively analyzed regarding (1) direction of the effect, (2) consistency of the effect across studies, and (3) strength of the evidence for the effect (evidence level, study quality) in accordance with the systematic review methodology [28, 34]. To identify structural and content components, the study reports of included studies were analyzed with a structured content analysis [35, 36]. Identification of structural components was directed by the question of how nurse counseling was performed (who counseled whom, where, when, how often, how long, with what, and under which counseling concept?). In contrast, identification of content components was driven by the question of what themes were addressed during nurse counseling (what was counseled?).Throughout the process, the category system (main and sub-components of nurse counseling) was developed by a deductive-inductive strategy. Main components were developed by deduction from the above mentioned questions, whereas sub-components were formed by induction and deduction (by consultation of similar work (e.g., [27, 37, 38]) and literature-based conceptualizations (e.g., [39, 40]). Coding was performed at the level of paragraphs, phrases, and words by progressively summarizing the material. Data synthesis is presented by a description of the category system that emerged, the frequency of the components, and their association with effects on specific patient outcomes under consideration of the best available evidence (by evidence level and study quality).

The literature search yielded 1356 records, from which ten studies could be included (Fig. 1). Seven investigations were experimental level 1 studies [25, 26, 41‐45] and three were pre-experimental level 4 studies [46‐48]. Three studies were of high [25, 43, 44], five were of moderate [26, 42, 45, 47, 48], and two were of low methodological quality [41, 46]. The interrater-reliability of the two authors who assessed the methodological quality was substantial (Cohen’s kappa = 0.84). Experimental trials suffered mostly from blinding of participants and an appropriate intention-to-treat approach. Pre-experimental trials mostly lacked strategies concerning confounding factors and an appropriate intention-to-treat approach (Additional file 3). The samples of three studies were considered as small, with fewer than 30 participants per study group [26, 42, 48] and therefore at risk for introducing type II errors. The studies by McCorkle et al. [43, 44] and by Donovan et al. [25] provide the best available evidence. Across all included studies, a total of 588 participants were investigated. Participants with various gynecological cancer sites were investigated most frequently [26, 41, 42, 45]. However, none of the studies included women with pre-cancer confirmed by colposcopy/biopsy. Across all included studies, 11 interventions were tested in total: the two reports of McCorkle et al. [43] and [44] address the same intervention, and each of the studies by Liu et al. [48] and Nolte et al. [45] examined two interventions. Two studies [45, 48] compared different counseling interventions against each other, while all others compared counseling with usual care. Of the ten authors contacted, six provided additional information. Study and intervention characteristics, effectiveness results, evidence level (thereinafter referred to as experimental or pre-experimental study), and study quality are presented in Table 1.

The effectiveness of nurse counseling is described in four main outcome groups: quality of life, symptoms, self-care performance, and healthcare utilization. The outcome group ‘quality of life’ refers to all reported effects of nurse counseling on quality of life total scores (such as overall quality of life) and quality of life sub-dimensions (such as cognitive or emotional quality of life dimensions) as measured and defined by the quality of life questionnaires used within the included studies. In distinction to the ‘symptom’ outcome group it is noteworthy that some quality of life measures used in the included studies contain questions about the occurrence of specific symptoms (e.g., [26]). For data synthesis it was not feasible to reconstruct which effects on symptoms were observed in these quality of life measures and to subsume them under the ‘symptom’ outcome group since studies did not report findings of single items. The outcome group ‘symptoms’ includes all reported effects of nurse counseling on physical and psychosocial symptoms outside quality of life measures as well as on the symptom severity, symptom distress, symptom consequences and symptom controllability of physical and psychosocial symptoms in general. In the outcome group ‘self-care performance’ only the performance of patients regarding physical symptoms was described in the included studies. Finally, the outcome group ‘healthcare performance’ refers to all reported effects of nurse counseling on the number of hospitalizations, oncology outpatient visits, emergency room visits, and primary care visits. The results are separated in each of these groups by experimental and pre-experimental studies.

In total, eight main components with various sub-components characterizing the structure of nurse counseling in gynecologic oncology were identified (see Table 2 that includes a definition for each coded component).

Data synthesis yielded four main components with different sub-components representing the themes discussed during nurse counseling (Table 2).

The quality of life results show less unequivocal evidence of good quality. The results that were replicated across studies (e.g., functional quality of life) are supported by weak evidence [26, 42, 46] and those with stronger evidence (e.g., overall quality of life) [43] are contradictory [26, 42]. The contradictory findings may be due to different quality of life measures used across studies. Some measures inquired about symptom occurrence (e.g., [26]), whereas others focused on the impact of symptoms on quality of life (e.g., [43]). Additionally, overall quality of life is based on single items in some measures (e.g., [26]) and on total scores in others (e.g., [42]). Another issue is that many study reports do not describe how an intervention may yield intended quality of life effects. Thus, few studies may have selected the most proximal measures of cause-effect relations. However, other studies of breast cancer patients also observed inconsistent effects of nurse counseling on quality of life [49‐51].

One study [48] showed that physical symptoms do not improve with counseling plus an add-on (leaflet, nurses’ position) when compared to standard counseling. Although this result is lacking strong evidence, it is reasonable that other factors contribute to the benefits of counseling. For example, previous research showed rather unfavorable results for the provision of written information alone [52]. The same applies to self-care performance, which also did not improve with counseling plus an add-on [48]. Overall, it is remarkable that only one study measured physical symptoms, although many interventions focused on their management [42, 43, 45‐48]. It is possible that these studies evaluated them within quality of life [42, 43, 46] and symptom severity [25] measures. For psychosocial issues, it is hardly possible to draw any firm conclusions since the evidence for them is almost always weak. Most robust results indicate that nurse counseling cannot improve depression [42, 43]. This may be because that depression is a state or a personality trait [53]. A non-psychotherapeutic intervention such as nurse counseling should primarily improve depression as a state. The measures of included studies [42, 43] do not make this distinction, and thus, possibly failed to show depression state changes. However, in other studies with breast cancer patients, depression also remained unchanged with nurse counseling [49, 50, 54]. The reported improvements of symptom severity and distress [25, 43], which relate to physical and psychosocial symptoms and rely on strong evidence, indicate that nurse counseling has the potential to impact biopsychosocial symptoms. Particularly regarding symptom distress, two studies, one that employed an APN intervention [43] and one that tested a web-based intervention [25], came to the same result. This promising result was also found in previous nursing research with breast cancer patients [55].

In addition, nurse counseling seems to have few effects on healthcare utilization [44]. McCorkle et al. [44] argue that their APN intervention reduced primary care visits because it addressed women’s everyday concerns, whereas emergency room visits remained unchanged because patients were encouraged to seek urgent care if necessary. Conversely, hospitalizations and oncology outpatient visits were scheduled follow-ups, which in cancer care rely on well-established, pre-defined protocols and were thus not preventable. In contrast, previous research showed reduced emergency room and hospital visits due to oncology nursing [56].

At least two-thirds of the interventions were implemented repeatedly [25, 26, 41‐46, 48], face-to-face [26, 41‐45, 47, 48], in an individual format [25, 26, 41, 43‐48], in the clinical setting [25, 26, 41‐48] with timely proximity to active treatment [25, 26, 41‐45, 47, 48] and thereby involved structured and tailored counseling [25, 26, 41‐44, 46, 47]. These components were also very common in previous work [27, 57] and are probably seen as part of a feasible, sustainable intervention structure. Around active treatment, patients have many clinical appointments, illness-related concerns, and self-care needs [19, 20, 22]. Their situation is mostly too complex and fluctuates too much for one consultation, and in the clinic, nurses can address patients’ questions after physician appointments. Of these rather frequently used components, repeated, individual consultations offered during and after active treatment, and delivered as both structured and tailored interventions, yielded the most robust improvements [25, 43, 44].

In contrast, few interventions involved home [26, 41, 43, 44], group [42], phone [42‐44, 46], or internet [25] counseling, additional family members [26, 43, 44, 47], or symptom assessment tools. The common absence of symptom assessment tools is noteworthy, since many interventions focused on symptom management [25, 26, 41‐48]. However, many of these components (e.g., phone counseling) can bridge critical transitions (e.g., discharge), between the clinical and home settings, but some might be difficult to implement (e.g., home counseling). Remarkably, these components were most often present in interventions with academically educated nurses [25, 26, 42‐44, 46], who are probably more familiar with sophisticated, innovative approaches and roles. In this context, it must be emphasized that this provider specification, as well as the utilization of symptom assessment tools [25, 43, 44], were among those components that were associated with the most trustworthy effects.

Most frequently, counseling addressed treatment-related physical symptoms [25, 42‐48], symptom etiology [25, 26, 41, 43‐47], symptom prevention and/or treatment [25, 26, 41‐48], and utilization of healthcare services [25, 26, 42‐46]. Nurses and patients may see these themes as being at the core of nursing practice and competence. Moreover, many interventions were carried out around the time of active treatment [25, 26, 41‐45, 47, 48], during which questions, such as what physical symptoms should be expected, and how they develop, how they are preventable and treatable, may have been more prominent than other themes. All these frequently used components were associated with significant effects in the studies providing the best available evidence [25, 43, 44].

In few interventions, counseling included disease etiology [42, 46], diagnostic or treatment-related procedures [26, 42, 46], treatment-related decision-making [43, 44], as well as goal setting and planning within symptom management [25, 43, 44]. Interestingly these themes were only discussed in interventions with academically educated nurses. We assume that disease and treatment-related issues (e.g., diagnostics) were not addressed in other interventions because they are traditionally more assigned to physicians’ competence. In particular, health-related decision-making can be an important theme when patients are in a dilemma to help them determine which options represent the most satisfying outcome for them [58]. However, in the most trustworthy studies, symptom-management was addressed comprehensively [25, 43, 44], whereas both the inclusion [43, 44] and exclusion [25] of health-related decision-making was associated with positive effects.

The results of this systematic review are limited to women with gynecologic cancer and our restrictive definition of pre-cancer excluded studies concerning less rigorously confirmed conditions. On the one hand, it was critical to draw the line at colposcopy/biopsy-confirmed conditions, and on the other hand, it was not reasonable to exclude women with pre-cancer from the research interest of this review since they need to be considered in gynecologic oncology. Furthermore, a broad concept of counseling was adopted. Thus, included interventions can also be characterized as psychosocial or supportive programs. Additionally, pre-experimental studies were eligible, even though these generally exhibit less internal validity than experimental investigations. The additional objective of identifying counseling components justified the sourcing of all available intervention studies using quantitative designs. Moreover, it was not feasible to reconstruct which effects on symptoms were observed in quality of life measures since studies did not report findings of single items. Partially limited study reporting also restricted the data synthesis regarding the structure and content of nurse counseling.