An integrative systematic review of employee silence and voice in healthcare: what are we really measuring?

1. Introduction

Healthcare organizations are unique, in the sense that the services provided involve risks that can range from minor nuisances to life-threatening and/or fatal consequences for patients, which puts a lot of pressure on healthcare professionals, staff, administrators, boards and policymakers. The reality of day-to-day practices in healthcare was brought into sharp focus globally during the recent COVID-19 pandemic, which highlighted the fragility of healthcare systems globally revealing the considerable stress experienced by healthcare staff (1). Both past and more recent inquiries into the failings of care have highlighted the critical role of communication and information sharing, indicating that speaking up and voicing concerns is an integral part of safe clinical practice (25). The same inquiries, however, have shown that (a) staff’s voiced concerns are frequently not acted upon until a disaster point is reached, (b) professionals with high calling intensity (i.e., professions with psychological contracts that encourage presenteeism even when employees are ill) frequently remain silent on critical issues related to patient safety and/or unprofessional behavior, and (c) whistleblowing is still considered the most “successful” channel to address systemic and organizational problems that have remained unresolved for a long time. Silence in health care has been related to concealing personal errors and covering errors made by others (6, 7), as well as reduced patient safety (8).

Two influential definitions of employee voice are those of Morrison (9) which describes “employee voice as informal and discretionary communication of ideas, suggestions, concerns, problems, or opinions about work-related issues, with the intent to bring about improvement or change” (p. 80) and that of LePine and Van Dyne (10) with voice being a term for “speaking out and challenging the status quo with the intent of improving the situation” (p. 853). More recently, the research on employee voice has been enriched by the increased interest in behaviors of withholding information from colleagues or superiors in the workplace, known as “employee silence”. One of the most influential definitions of employee silence has been provided by Pinder and Harlos (11): “an employee’s intentional withholding of genuine expression about behavioral, cognitive, and/or affective assessments of organizational conditions to organizational members who seem capable of changing the situation”. Tangirala and Ramanujam (12) define employee silence as “employees’ intentional withholding of critical work-related information from other members of their workgroup” (p. 41). However, it remains unclear what is included in the terms “employee voice” and “employee silence” in healthcare, as they can sometimes be discussed in terms of safety voice and safety silence (i.e., speaking about patient safety/patient advocacy or concealing information related to patient safety, respectively); how employee voice/silence fits within the theoretical literature on organizational culture and behavior in healthcare; and/or whether it extends to the professional culture and identity of healthcare staff. This is also related to the fact that although there is a general agreement that employee silence refers to withholding information and employee voice refers to sharing information in the workplace, any attempt to operationalize employee voice/silence reveals difficulties and challenges in identifying what should be considered voice/silence. This can in turn affect the way voice/silence are measured. For example, voice is defined as a discretionary behavior, in that individuals choose whether to engage in verbally expressing themselves or not at any particular moment with this being affected by a variety of factors (9). Similarly, the definitions of silence presented previously define silence as the “intentional withholding”. In healthcare though, the notion that voice/silence is a discretionary behavior can be easily misinterpreted, especially if we take into account the type of information that is often likely to be conveyed in healthcare: a concern related to patient safety and/or quality of care. Thus, the content and the context of speaking-up can differentiate the extent to which silence or voice are discretionary, as concealing a medical error for instance has ethical, moral and legal ramifications.

Recent literature has suggested that, although on a lexical level employee silence and employee voice seem to be opposite terms, they might also be distinct concepts with different antecedents (13). Moreover, silence and voice can occur at the same time, meaning that employees might be speaking up in some situations (or regarding specific issues), but withholding their voices in other situations. For example, the definition by Pinder and Harlos (11) specifies the withholding of “genuine expression”; this means that even in situations where employees engage in speaking, it cannot be ascertained that they are not engaging in any form of withholding voice (e.g., instead of speaking up about the unprofessional behavior of a colleague they may share a generally neutral comment on workplace behavior). This is particularly relevant to healthcare organizations, where a significant amount of the information shared (or withheld) is frequently related to patient safety and quality of care, which involves the interests not only of the healthcare professionals and the organization itself, but also those of the patients and their families—which has also been discussed as a conflict of interest (14).

The increasing empirical evidence regarding speaking-up in healthcare suggests that silence is the norm while voicing concerns is met with negative consequences for employees (1518). For example, employee silence has been linked to employee well-being in the literature (19). In terms of understanding how silence/voice links with different outcomes for employees, we build upon the example of employee well-being, and more specifically burnout (19). The Job Demands-Control Model (JD-C) (20, 21), the Job Demands-Resources Model (JD-R) (22, 23) and the Conservation of Resources Model (COR) can help advance our understanding of employee voice behaviors in healthcare organizations (24, 25) and their links to employee outcomes. Both the JD-C and JD-R models view burnout through the lens of a mismatch between demands and resources; in this context employee silence could be evidence of the mismatch while employee voice could be evidence of a better fit. COR emphasizes the tendency of individuals and groups to always aim to obtaining, retaining, fostering, and protecting the resources they centrally value. One of the main principles of the COR theory suggests that when employees’ resources are (almost) depleted, individuals are more likely to enter a defensive mode to preserve the remaining resources or to seek for alternative survival/adaptation strategies if previous experiences were found to be maladaptive and consuming; these defensive modes can be sometimes aggressive and/or irrational. A common response, for example, might be defensive withdrawal, allowing the individuals to gain time to regroup, wait for help and allow the stressor to pass (26). Viewed through this lens, silence could be the result of an employee moving into a defensive mode in response to depleted resources.

Greater understanding of how employee voice/silence among healthcare professionals is conceptualized and measured is proposed as a potentially effective way to identify what is considered employee voice/silence in healthcare. It has been argued that withholding concerns is the norm in healthcare (27) and while healthcare organizations might share some common antecedents of voice behaviors with other industries (e.g., fear of retaliation or losing one’s job; not wanting to risk relationships with colleagues etc.), there are specific characteristics of healthcare education and professional culture that should be taken into account when examining voice among healthcare workers (28). Meta-analytic findings suggest that interventions to improve speaking up in healthcare achieve disappointing results (29). In order to better understand why this is happening, it is necessary to understand how employee silence and voice are operationalized in healthcare and whether there is a need for a new framework adapted for healthcare organizations specifically. By synthesizing and/or critiquing existing research on employee voice/silence among healthcare professionals, an Integrative Systematic Review can offer new insights and new ways of understanding the phenomenon (30). Therefore, the purpose of this integrative review was to explore the following questions;

(1) How are employee voice and employee silence conceptualized and measured in healthcare?

(2) What is the theoretical background to employee voice and silence in healthcare?

2. Method 2.1. Methodology

The methodology used in the present review involved an integrative systematic literature review. Reviews should meet the same standards of methodological rigor as primary research (31) and given that it is complex to combine various methodologies into one review it becomes even more important to use an explicit and systematic method to avoid inaccuracy and biases (32). Conventional systematic reviews and meta-analyses are the preferred methodology when the available data is appropriate. However, when a phenomenon requires clarification and insight, which involves a more interpretive synthesis of existing literature, an IR which combines the best elements of a systematic review and narrative review is more appropriate. According to Greenhalgh et al. (33) the systematic review format has been erroneously defined as a universal gold standard, partly because the term “narrative review” is frequently misunderstood, misapplied and unfairly dismissed. Toronto and Remington’s (34) six steps method served as a framework for this Integrative Review (IR). The six steps are: (1) problem formulation, (2) data collection via systematic literature search, (3) evaluation of data points by analyzing quality and relevance of selected literature, (4) data analysis and interpretation, (5) presentation of results via discussion and conclusion and finally, and (6) dissemination of findings. This framework is built on Cooper’s (31) approach for conducting IRs with a transparent and rigorous systematic approach to reviewing the literature.

2.2. Problem formulation

There is a gap in our understanding of employee voice and silence in healthcare, and the relationship between withholding information and healthcare outcomes (e.g., patient safety, quality of care, worker wellbeing) is complex and differentiated. While there is empirical evidence suggesting some understanding of certain aspects of voice/silence among working populations in general (e.g., the relationship between employee silence and employee well-being) (19), the uniqueness of the healthcare sector means that a tailored approach is needed to further understand what drives voice/silence in healthcare; how it impacts quality of care; and what factors need to be addressed in order to identify suitable and relevant practical implications. This highlights the needs to review what measures are used to assess employee voice/silence in healthcare as well as whether the existing research is driven by distinct theoretical frameworks. The complexity of investigating the phenomena of employee silence and employee voice in healthcare is suited to the holistic approach of the IR. According to Toronto and Remington (34), the IR approach looks more broadly at a phenomenon of interest than a systematic review and allows for diverse research, which may contain theoretical and methodological literature to address the aim of the review. Moreover, the IR approach supports a wide range of inquiry, such as defining concepts, reviewing theories, and analyzing methodological issues.

2.3. Data collection via systematic literature search

A review protocol was registered on the PROSPERO register (CRD42022367138). An electronic database search for the period from January 2016 to January 2022 was conducted by two researchers (OL and MKJ), independently. Librarians were also consulted throughout the search process to ensure its quality (34). The following electronic databases were used: PubMed, PsycINFO, Scopus, Embase, Cochrane Library, Web of Science, CINAHL and Google Scholar. Additionally, analysis of references lists of retrieved studies and manual scoping was conducted. Given the large amount of literature on the subject and the need to analyze the most recent publications, the search was limited from January 2016 to January 2022. This condensed time-period was wide enough to provide an appropriate snapshot of the literature and dense enough to examine the range of conceptualizations, measures and potential implications in the literature. Initially the search period was set at 10 years, but the amount of research papers became unwieldy making the review process unmanageable, so a decision was made to analyze the most recent papers in the last 6 years. The choice of 6 years was based on the need to capture a manageable number of papers, as well as the following factors: (a) the amount of research around employee silence/voice in healthcare has peaked since 2016 and even more so after the COVID-19 pandemic; (b) the way that employee silence/voice has been measured in healthcare has been recycled over the last 10–15 years.

Keywords used in the systematic search were identified by reviewing article examples and results found in preliminary searches (34). The keywords identified were then put together into a search string (for the search strings used for the different databases see Supplementary material 1); for example:

(“Employee silenc*” OR “Employee voic*” OR “Organizational silenc*” OR “Organisational silenc*” OR “Organizational voic*” OR “Organisational voic*” OR “Silence behaviour” OR “Voice behaviour” OR “Voice behavior” OR “Speak up” OR Speak-up OR “Prohbitive voic*” OR “Promotive voic*” OR “speak-up related climate” OR “silence culture” OR Concealment OR “Truth disclosure” OR “Transparency” OR “Error concealment” OR Confidentiality OR “medical error*” OR “rais* concerns” OR Whistleblow* OR Whistle-blow* OR “blowing the whistle”)AND (Healthcare OR “Health care” OR “health organization” OR “health organisation” OR “health service” OR “medical service” OR hospital OR hospitals OR “primary care” OR “healthcare employee*” OR “health care employee*” OR “health personnel” OR “health employee” OR nurs* OR physician* OR doctor* OR medic* OR “patient safety”)

The search included both title and abstract. The benefit of the IR approach is that it creates conditions for a comprehensive search of literature (30) and a broad search was conducted. Five inclusion criteria were applied. These were: (1) the paper had to be in the English language, (2) the paper had to be primarily based on quantitative research, (3) only peer reviewed journal papers were included, (4) voice and/or silence had to be measured quantitatively, either directly via silence or voice scales or indirectly via other scales incorporating voice or silence constituents, and (5) the search was restricted to employees in healthcare organizations. Papers were excluded if 1) the full text was not in the English language, 2) there was no measure of either employee voice or employee silence and 3) the sample did not consist of healthcare organization employees. The data retrieved from each database was logged in Excel and a duplicate control was conducted using EndNote. The remaining data was imported into Rayyan where two reviewers (OL and MKJ) had activated the blinding function in order to not see each other’s decisions, labels and notes.

The results from the two independent reviewers (OL and MKJ) were compared (Cohen’s k = 0.96) and differences were discussed until an agreement was reached. A third reviewer was available when needed (AM). From the 209 studies that were screened in full text, a total of 76 papers met the criteria and were eligible for inclusion in the review. The PRISMA diagram below illustrates the number of papers identified, included and excluded throughout the process. In the first screening step, the abstracts were retrieved and read in order to decide which papers were relevant to include, based on the identified inclusion criteria. A full text review was conducted for the second screening step, using the same criteria, and the reasons for exclusion at this step are presented in the PRISMA diagram (see Figure 1). The most common reason for exclusion was that the study did not measure voice or silence.

www.frontiersin.org

Figure 1. PRISMA flow diagram.

Data were extracted from the 76 studies regarding the following information: Lead author and date; Study location; Professional group (e.g., nurses, physicians); Study design (e.g., cross-sectional/longitudinal; correlational/experimental); Sample size at baseline and follow-up; Female %; Measure of employee silence and/or voice outcome (see Table 1).

www.frontiersin.org

Table 1. Studies included in the review.

2.4. Quality appraisal

Quality assessment was conducted using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (59). Three reviewers independently assessed the quality of the included studies (CM, KP and CC) and a fourth reviewer was invited in case of disagreement (OL or AM). The tool contains 14 criteria, and the evaluator is asked to answer whether the study in question meets the criterion, with the possible answers being “Yes, No, Cannot Determine, Not-applicable, and Not Reported”. A score of >/=11 corresponds to good quality, 7–10 to fair quality and < 7 to poor quality. As this was an Integrative Systematic Review employing a narrative synthesis, risk of bias assessment by generating a synthesized result (e.g. a meta-analytic effect estimate, or median effect across studies) was not possible, meaning that the process is more conducive to systematic reviews/meta-analyses in which the sampling frame is narrow, the research designs included are similar or even identical and outcomes assessed are not fundamentally heterogeneous (32).

2.5. Data analysis and interpretation

As this integrative review looked more broadly at the phenomenon of employee silence and voice in healthcare, with the aim of creating clarifications and insights, a holistic approach was sought. No papers were excluded based on quality assessment results, even though such a quality assessment was conducted. We summarized the data extraction results using descriptive statistics.

To supplement the findings regarding the measures of employee voice and silence in healthcare, thematic analysis of the relevant measures of each study was used to identify themes with constant comparative analysis (47, 60). A constant comparison method is a widely used approach used that allows the conversion of data into systematic categories, which allows the researcher to identify distinct themes or variations. In IRs, this approach is compatible with the use of varied data from diverse methodologies (32). The coding process was inductive, meaning that there were no pre-defined codes and an extensive review of the contents of each measure was conducted to develop initial codes. It is a common challenge in qualitative analysis that initial coding leads to an overwhelming number of shallow codes (47). Discussion among three authors (OL, MKJ and AM) helped develop the analysis throughout the process as multiple instances of codes occurred in close proximity to each other highlighting potential connections between codes. Each study was independently analyzed and coded by OL in conjunction with one other independent reviewer (MKJ, AM, CM, KP). The coding process was reviewed until agreement was reached regarding the most meaningful criteria and categories of analysis emerging from the process. In a process of constant comparison, extracted data on measures were compared item by item so that similar data were categorized and grouped together, and these coded categories were then compared to further the analysis and synthesis (32). OL collated all proposed codes, which the research team collectively discussed to create a comprehensive and shared understanding of each code. Due to the diverse representation of employee voice and silence measures, these were coded using constant comparative analysis according to the following criteria: a) context, meaning whether the measures were patient-safety specific or aimed at measuring employee silence/voice in a general context; b) conceptual distinction, meaning whether data on employee silence (withholding voice) or employee voice (speaking up) were collected; and c) aspects, meaning what aspects of employee silence and/or voice/speaking up the measures aimed to capture. Based on the first criterion (a) two meaningful categories were coded as 1) patient-safety specific and 2) general context. For the second criterion (b), measures were coded as 1) measures of employee silence, 2) measures of employee voice/speaking up and 3) measures of multiple aspects of employee silence and employee voice/speaking up. For the third criterion (c), measures were coded based on what specific aspect of employee silence/voice they were aiming to capture; examples include: antecedents of silence (i.e., motives and content); intention to speak up; perceptions of speaking-up related climate; self-reported past speaking up behavior; externally observed speaking-up behaviors. All authors reviewed and agreed on the themes’ codes from the data analysis. The themes and subthemes are presented in Table 2.

www.frontiersin.org

Table 2. Categorization of studies according to context, conceptual distinction, and aspects of employee voice/silence.

3. Results 3.1. Sample

The total sample of participants was 122,009 (N = 122,009, 69.3% female), while 13% of the included studies did not report information of the participants’ gender (40, 54, 6165)). Of the total sample, 57,520 were identified as nurses (47.1%); 11,228 (9.2%) were nurse managers/ supervisors/ head nurses; 1736 (1.4%) were nurse experts; 9,266 (7.6%) were physicians; 4,279 (3.5%) were residents or trainees; and 392 (0.32%) were senior physicians. Moreover, 18,468 (15.1%) were allied health professionals (e.g., physiotherapists, psychologists, dieticians etc.); 5,792 (4.7%) participants were classified as “other” and 132 (0.11%) as administrative staff. Seven studies had unspecified samples (n = 2,609; 2.1%): Avgar et al. (49) with 363 unspecified healthcare workers; Gupta and Ravindranath (66) with 1700 unspecified eye hospital employees; Hu and Casey (67) with 165 unspecified health care workers; Lawson et al. (64) did not report sample size of participants (perfusionists); Lemke et al. (68) did not specify on their sample of 49 anesthesia care providers; Mesdaghinia et al. (69) reported a sample of 203 supervisors in a hospital; finally, Roussin et al. (70) did not provide the numbers of nurses and physicians in their sample (n = 129). Thus, based on the information provided, most of the participants were nurses (57.7% cumulative percentage of nurses, nurse managers and nurse experts).

Detailed information on departments/units was available only for 13.5% of the total sample. Based on the information reported in the studies, 7.2% of the participants worked in surgical departments; 2.9% in Anesthesiology; < 1% in ICUs; 1.4% Eye Hospital; 1.1% in Internal Medicine; 0.4% in Emergency Departments, followed by Pediatrics (0.3%); Cardiology (0.2%); Radiology (> 0.1); OB/Gyn (0.1%); Nephrology (0.07%); Gastroenterology (0.07%); Orthopedics (0.07%); OR (0.03%); Outpatient Units (0.02%); Plastic Surgery (0.02%).

In terms of location, 21 studies were conducted in United States (26.9%)—one study was conducted in both Japan and the United States (1.3%)—10 studies were conducted in Turkey (12.8%), eight in Switzerland (10.3%), five in China (6.4%), four in Egypt (5.1%), three in South Korea (3.5%), three in Australia (3.5%) and three in the Philippines (3.5%); in two studies, participants were recruited form the Netherlands (2.6%) and in four studies participants were recruited from Germany (5.1%). Two studies were conducted in Jordan (2.6%). One study (1.3%) was conducted in each of the following countries: Austria; Spain; Indonesia; Saudi Arabia; Ireland; Greece; Pakistan; Cyprus; Canada; New Zealand; Taiwan; India; UK; Iraq; Iran. Thus, the most represented country was the United States with 21 studies (26.9%); however, it was noteworthy that 47.8% of the studies included in the current IR were conducted in countries outside of Europe, the United States and North America.

In terms of publication year, 19 studies were published in 2021 (24.4%), 14 studies in 2020 (17.9%), 10 studies in 2019 (12.8%), 15 in 2018 (19.2%), 12 in 2017 (14.5%), nine in 2016 (11.5%).

3.2. Quality appraisal

Of the 76 research reports that were included in the current integrative review, 8 (10.5%) were rated by three independent evaluators as good, 36 as fair (47.4%) and 32 studies (42.1%) were rated as having poor quality on the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies. This is strongly linked to the fact that the studies were predominantly cross-sectional, did not provide justification for the sample size (e.g., prospective power analysis) and did not always control for the effect of important relevant variables (e.g., confounders).

3.3. Measures

In total, 45 distinct measures of employee silence and employee voice/speaking up were identified across the 76 studies included in this IR. Thirty of these measures were identified as safety-specific, of which three were measures of employee silence, 23 were measures of employee voice/speaking up and four were measures including multiple aspects of patient safety-related silence and/or voice/speaking up. Of the remaining 15 general context measures, seven were measures of employee silence and 8 were measures of employee voice/speaking up. Examples of items for each measure are included in Table 3. Employee silence was measured in 31(40.8%) of the 76 studies and employee voice was measured in 53 (69.7%) out of the 76 studies, as some studies measured both voice and silence together.

www.frontiersin.org

Table 3. List of employee silence and voice measures.

3.3.1. Patient-safety specific measures

The majority of distinct measures—30 out of the 45—identified across the studies included in the current review were safety specific. This category includes measures that are focused on safety issues.

3.3.1.1. Patient safety employee silence antecedents

Three distinct measures were identified in this category across five studies. All three measures aimed at capturing antecedents of employee silence regarding patient safety issues (e.g., motives). Two of these measures were developed to measure motives for patient-safety silence and each was used in one study, and one for patient-safety related silence in specific situations/regarding specific content.

A scale for self-reported frequency of remaining silent about patient safety due to specific reasons was developed and tested by Abd El-Fattah Mohamed Any et al. (61), examining the following motives: avoidance, belief, attitude, fear, management and organization.

The “Safety Silence Motives” scale was developed and tested by Manapragada and Bruk-Lee (65) to measure frequency of remaining silent for reasons that are relationship based, climate-based, issue-based and job-based motives. Both (61, 65) were validation studies.

The third measure in this category is the scale adapted by Tangirala and Ramanujam (12) and was used in three (3.9%) of the 76 studies (18, 48, 71).

3.3.1.2. Likelihood/ intentions of speaking-up regarding patient safety

Six distinct measures were identified in this category across seven studies. The measures in this category aimed at capturing the likelihood/intentions of speaking up about patient safety.

Three of the measures included in this category were questionnaires that used hypothetical scenarios (vignettes) related to patient safety issues. These include: the Speaking-up Scale developed by Andrew and Mansour (72) which was used in the study of Mansour et al. (45); a questionnaire with four hypothetical scenarios (73); and the study of Lowenbruck et al. (51), evaluating participants’ disclosure intention using two hypothetical scenarios with increasing adverse outcome severity.

The remaining three measures in this category were scales asking the participants to rate their speaking-up intentions. The Individual Speaking up Attitudes was used by Alingh et al. (44), which was an adaptation of the Communication Openness Scale (74). The voice subscale (four items) from the Safety Citizenship Behavior Scale (42) was used by Hu and Casey (67) to measure Safety Voice. Although the scale refers to safety in general (not specifically for healthcare/patient), it is included in the patient-safety specific measures, as in healthcare any reference to safety is always linked to patient safety. Best and Kim (75) measured participants’ likelihood of speaking up about patient safety pre- and post-intervention using two items (towards a team member and towards persons who are capable of change).

3.3.1.3. Speaking-up about patient safety climate

The measures in this category include items that aimed at capturing the participants perceptions regarding speaking-up-related climate for patient safety issues. Ten distinct measures were identified in this category across 16 studies.

The Hospital Survey on Patient Safety Culture (HSOPSC) developed by the Agency for Healthcare Research and Quality (AHRQ) (41) was used in six (7.9%) of the 76 studies (50, 63, 64, 7678). The survey includes four items related to speaking up about patient safety in the “Communication” subscale.

The Safety Attitudes Questionnaire (SAQ) (79) was used in three studies (3.9%) (62, 80, 81) which includes items relevant to speaking-up in the Teamwork Climate and the Safety Climate subscales.

Each of the remaining eight measures was used only in one study. In particular, a modified version of the Teamwork Climate Survey (82) was used by Ginsburg and Bain (83). The Speaking up Climate for Safety Scale (88) and the Speaking up Climate for Professionalism Scale (88) were used in the study of Martinez et al. (80).

One item related to the speaking-up climate from the Safety Climate Survey (84) was used in the study of Gauld and Horsburgh (85). Avgar et al. (49) measured Employee Patient-Care voice using two items adapted from Clark et al. (89), assessing the degree to which employees perceived that their input regarding important patient care issues is taken into account. Ng et al. (87) measured the participants perceptions of speaking-up using a scale developed to measure communication openness based on the work of Reader et al. (86), consisting of separate questions for nurses and doctors. Jungbauer et al. (91) included modified items from the work of Wu et al. (90), to measure “reporting-specific trust” which are similar to those used in other scales to measure perceptions of speaking up related climate.

The Safety Communication Scale (39) was used in one study (65) to assess how comfortable participants felt in sharing their concerns and ideas regarding safety. Although the scale refers to safety in general (not specifically for healthcare/patient safety), it is included in the patient safety specific measures, as for similar reasons to Hu and Casey’s (67) Safety Voice mentioned above.

3.3.1.4. Externally observed speaking-up behaviors

This category includes measures that aimed at capturing aspects of externally observed speaking up/voice behaviors. The following seven distinct assessments were used to rate externally-observed speaking-up behaviors across seven observational studies: the Advocacy-Inquiry rubric in the study of Guris et al. (92); the Co-ACT coding system in two studies (93, 94); an observation system based on both organizational behavior and anesthesia research (68); the five-point modified Pian-Smith (95) grading scale ranging from 1 being silence to 5 being repeated inquiry (96); level of speaking-up during the scripted opportunities when speaking up was anticipated in two studies (92, 97); assessment based on predefined desired actions for each event (97); and frequency of observed behaviors that were identified through focus groups as encouraging speaking up (e.g., restate; ask open questions; clarify, etc.) (65).

These seven observational studies (9.2% of the 76 studies)—meaning that speaking up/voice was measured by observing participants’ speaking-up behaviors—also discussed the following aspects of speaking-up: voicing frequency (68, 93), the level of assertiveness (68, 96), content (e.g., patient-safety concern, innovative ideas) (68, 93), and who the participants spoke-up to (93, 97). In some cases, other aspects were taken into account as well, such as clinical relevance (68), whether speaking up was prompted or unprompted (92), quality (92), and time to voice (93). One study was focused on the observation of behaviors encouraging speaking up rather than monitoring actual speaking-up behaviors (54).

3.3.1.5. Scales measuring multiple aspects of patient-safety silence/voice

This category includes four measures that aimed at capturing multiple aspects of silence and/or voice/speaking up included in one questionnaire. There four measures were identified across 11 studies.

The first measure in this category is the Speaking Up about Patient Safety Questionnaire that was used in six (7.9%) of the 76 studies included in this IR (55, 78, 98100, 131). The measure includes one subscale that measures the frequency of having remained silent regarding specific content, conceptualized as withholding information about patient safety; one subscale that measures the frequency of having spoken up regarding specific content; and four questions measuring the likelihood of speaking up in a hypothetical situation.

Martinez et al. (80) used a questionnaire related to multiple aspects speaking up behavior, including past speaking-up behaviors; antecedents of speaking up about patient safety or unprofessional behavior (facilitators and barriers); speaking-up related climate and the likelihood of speaking up towards different team members (e.g., nurse, intern) with regard to two scenarios. Kesselheim et al. (102) also used this questionnaire.

Ortiz-Lopez et al. (101) developed and tested a scale focused on speaking up about medical errors, including subscales for attitudes towards speaking up; perceptions of speaking-up related climate; intention to speak up; and past speaking up behavior.

Toy et al. (52) used a questionnaire that measured the likelihood of speaking-up conceptualized as assertive communication, with three hypothetical scenarios related to patient safety concerns and 10 items measuring speaking-up antecedents, including intrapersonal factors that might influence speaking-up specifically for the operating room. The latter was also used in the study of Guris et al. (92). An interesting aspect of the latter is that the authors included potential positive outcomes of speaking up (e.g., “Speaking up in the OR will increase my colleagues’ respect of my patient care skills”) as opposed to negative outcomes which is more common in voice/silence scales.

3.3.2. General context measures 3.3.2.1. Employee silence antecedents and consequences

Seven distinct measures were identified in this category. Five of these measures aimed at only measuring antecedents of employee silence, while the remaining two measures aimed at capturing antecedents and consequences of employee silence.

The most frequently used scale to measure employee silence motives in this category was the scale developed by Van Dyne et al. (53), which appeared in 9 (11.8%) out of the 76 studies included in this IR (17, 3638, 103107). The Van Dyne et al. (53) scale measures three types of employee silence based on three distinct motives: prosocial silence, acquiescent silence, and defensive silence. Al-Abrrow (103) re-labelled two of the three types of silence: quiescent silence (instead of acquiescent) and positive social silence (instead of prosocial silence).

The Yalcin and Baykal (40) scale for employee silence antecedents was used in two studies (2.6% of the 76 studies) (40, 108). This scale required the participants to indicate the frequency of remaining silent due to reasons related to silence climate, silence based on fear, acquiescent silence and silence based on protecting the organization. Though the scale was tested for use among healthcare professionals, it is included in the general context category as the items do not refer to patient safety specific silence.

The third measure in this category was developed by Alheet (43). The ad-hoc questionnaire purports to measure causative factors in the following dimensions: management and organization; experience; anxiety and fear; and being afraid of alienation. Though the scale is initially presented to be a multidimensional measure of causative factors, the authors analyze it as a unidimensional variable.

The fourth measure in this category was that developed by Jain (109), which was originally constructed to investigate dimensions of employee silence in the Indian work settings and specifically focusing on the supervisor-subordinate relationship; the scale was used by Mousa et al. (110) to measures reasons for employee silence with respect to their supervisors, organized in four factors: fear of retaliation; internal motivation; self-competence; self-image. The model was treated as a one-factor-model in this study (110).

The fifth measure in this category was developed by MacMahon et al. (112) to assess bullying reporting antecedents and was based on the work of Pinder and Harlos (11) on the antecedents of employee silence, as well as the work of Whiteside and Barclay (111).

As mentioned earlier, two scales aimed at measuring both antecedents and consequences of employee silence. The scale developed by Vakola and Bouradas (46) was used in one study (65). The last scale in this category—and the second to measure both antecedents and consequences—was the employee silence scale developed by Cakici (5658) that was used in five (6.6%) of the 76 studies included in this IR (35, 113116). Antecedents include items related to both motives and content. The consequences subscale was used only by Seren et al. (117). Caylak and Atluntas (114) only measured the antecedents and the other three studies (35, 113, 115) only measured the frequency of silence motives.

3.3.2.2. Employee voice/speaking up antecedents

Two distinct measures of employee voice antecedents (e.g., motives, content) were identified across 12 studies out of the 76 included in this review.

The Voice Behavior Scale by Van Dyne and Lepine (117) was used in six (7.9%) out of the 76 studies (56, 66, 107, 119121). Items refer to speaking up in the team as a promotive behavior—also defined as prosocial or promotive voice (30)—and are based on the approach that views voice as one of the helping and/or extra role behaviors (117).

The employee voice scale developed by Liang et al. (122) was used in six (7.9%) of the 76 studies (69, 123126). This scale measures promotive voice and prohibitive voice where the former refers to “putting forward new ideas and methods to improve the efficiency of the enterprises” and the latter refers to “expressing the inhibitive viewpoint and the harmful problem that hinders the efficiency of the organization”. Mesdaghinia et al. (69) used the shortened version to measure prohibitive voice (122).

3.3.2.3. Employee voice/speaking-up self-reported behavior

Two distinct measures of self-reported past employee voice behavior were identified across two of the 76 studies included in this review. The Upward Voice Behavior scale by Liu et al. (132) was used in one study (127). The scale consists of three items requiring the participants to indicate the frequency with which they voice their opinions and concerns to their supervisors. The second was an ad-hoc questionnaire used by Roussin et al. (70) asking the participants to indicate how often they spoke up in a training program.

3.3.2.4. Speaking-up related climate

The measures in this category include items that aimed at capturing the participants perceptions regarding speaking-up-related climate. Four distinct measures were identified in this category across 16 studies.

Bilotta et al. (128) used an ad-hoc scale consisting of three items to measure what they refer to as “organizational-level employee voice”. The three items were related to the perceived organizational climate allowing for voice, initiative and autonomy. Carpini and Flemming (129) reported that they measured speaking up as part of their “Team Behaviors” scale, asking the participants to indicate the extent to which they observed speaking-up behaviors and/or a reluctance to speak up (silence), among other aspects of team behavior. Ridley et al. (77) in their study examining the effectiveness of a teamwork training asked the participants to indicate after each surgical case whether they felt comfortable asking questions and expressing concerns and whether medical errors were reported. Finally, Holland et al. (130) operationalized their approach to voice as that of the “direct voice” defined as occurring in specific two-way communication channels, thus measuring the existence of formal voice channels within the organization.

3.4. Theoretical approaches to employee silence and voice

The majority of papers reviewed provide relatively little detail on the theoretical background behind the authors’ approaches to employee voice and silence. Given the paucity of discussion concerning theory construction and development among the papers, simply highlighting these gaps would not be informative or useful. Instead, we have focused on the papers that have attempted to explore theoretical aspects in detail.

Bilotta et al. (128) linked a theory of organizational justice known as the group engagement model (GEM) (129) to employee voice and psychology safety in healthcare. They found robust support for their theoretical model in the healthcare context and found that employee voice accounted for 39% of the variance in the relationship between organizational-level fairness and patient mortality. From a theoretical perspective, the research identified areas for further investigation concerning psychology safety. For example, they found that providers may witness unsafe behavior by their colleagues or supervisor, which in turn leads employees to develop a shared perception of when it is worthwhile to speak up and rely on their fairness perceptions as an indicator that speaking up will not be met with unjust repercussions or unreasonable sanctions. Hu and Casey (67) analyzed employee voice via its theoretical links with social identity theory, organizational identification and psychological safety. They found evidence of a complex interaction between safety motivation, psychological safety and safety voice. Specifically, they found that the relationship between safety motivation and safety voice was only significant when psychological safety was low or at an average level. The authors were unable to adequately explain the complex interaction, but they suggest that organizational identification is worthy of more theoretical consideration given that studies have reported that psychological safety can be very low among individuals working in nursing (133). The aforementioned papers should prompt us to consider the speculation of Roussin et al. (70) that “psychological safety microclimates” can help explain the variation in the results regarding the relationship between employee voice/silence and psychological safety. The idea of ‘microclimates’ fits with the suggestion that healthcare organizations are composed of multiple smaller organizations.

Kaya and Bacaksiz (107) examined the relationships between nurses’ positive psychological capital, and their employee voice and organizational silence behaviors. The paper reviewed in detail the concept and theory of positive psychological capital (PsyCap). PsyCap is in line with what is considered to be Positive Organizational Behavior and has four components: self-efficacy, optimism, hope, and resilience (107). Interestingly, the study found that the nurses with higher education levels had lower PsyCap levels, and that nurses with high PsyCap levels remained less silent for individual reasons, but relational reasons cause them to remain silent more. The authors did not adequately explain their contradictory findings, but it further echoes Hu and Casey’s (67) findings concerning the importance of organizational identification. Kesselheim et al. (55) on United States Pediatric trainees provided no information on the theoretical background to their research. However, their results dovetail with aforementioned implications regarding organizational and professional identity (67, 107). Their research on voice among United States Pediatric trainees at 2 large US academic children’s hospitals indicated that, while more than half of the respondents reported observing unprofessional behavior during their most recent inpatient month, strikingly few (20%) respondents anticipated speaking up to an attending in the case of unprofessional behavior, even when they perceived a high risk of patient harm. The observed deficits in speaking up were even more notable considering the vast majority of participants reported prior training specifically related to speaking up. The authors recommend that such stark results call for more assertive communication skills training and anonymous reporting procedures. However, this paper highlights the dangers inherent when there is a lack of any substantive consideration of the drivers (theory) of silence and voice, resulting in recommendations that are generic and cosmetic that cannot be linked back to a suitable evidence base. Research that has closer links between theory, methods and outcomes provides more nuanced outputs. For example, Krenz et al. (93) examined team composition and interprofessional teamwork among healthcare providers and found that stronger hierarchy and more centralized leadership delayed nurses’ voice but did not affect the overall frequency of voice. The researchers directly linked their theory (i.e., contrasting how individuals think they would act and how they actually act) with their method of investigation (i.e., simulation scenarios). Additionally, Loewenbruck et al. (51) studied voice expression among physicians in Germany, Japan and the United States. Their research reviewed the theoretical links between intrinsic and prosocial employee motivation, role-model-based learning power distance and the Theory of Planned Behavior. Contrary to their expectations, internal attribution of medical error prevented voice. The paper is an excellent example of why extensive discussion of theory leads to clearer testable hypotheses that can produce unexpected results, and that contributes to our theorizing about employee voice/silence. Mesdaghinia et al. (69), in a sample of hospital employees working in clinical and administrative positions, connected prohibitive voice with leader-member exchange, moral identity, and moral symbolism. The authors provided an extensive discussion of the theory of moral identity, and specifically assessed moral identity internalization and symbolization. Both these concepts link with the common themes throughout the papers of professional identity and the organizational environment (i.e., organizational identity, psychological safety, organizational level fairness). The reviewed papers generally mentioned classic theories in organizational behavior and occupational health (e.g., Conservation of Resources model, the Job-Demands-Resources model, leadership member exchange theories, social exchange theories and social identity theories) without detailed explanation of how their research links with existing theoretical models and approaches to employee silence/voice. Only one paper approached employee silence/voice directly from an industrial relations perspective. Avgar et al. (49) provide a detailed review of the literature in Labor-Management Partnership (LMP) initiatives, which include direct and indirect methods for eliciting employee input, involvement, and voice. Their research sought to understand why industrial relations approaches have tended to have limited impact on voice. In their research, patient-care voice was enhanced by the positive relationship between quality of LMP processes and employee trust, but the effects of partnership were a function of more than merely participating. In both ANOVA and regression analyses they did not find a significant relationship between participating in an LMP initiative and any of the other variables in the study. Thus, we are left with the unanswered problem of why union effectiveness was not important.

4. Discussion 4.1. Measures of employee voice/silence

Overall, the 76 studies included in the current IR indicated a high level of heterogeneity regarding the measures of employee voice and silence in healthcare as well as the conceptual descriptions of voice/silence in healthcare. We identified 45 distinct measures, the majority of which was safety-specific and related to employee voice/speaking up. One reason for this heterogeneity could be that different definitions of the concepts are used. This lack of theoretical grounding is directly linked to challenges related to the validity of measures. For one, different theoretical approaches to voice/silence generate different operationalizations, which in turn lead to collecting heterogenous data while claiming to measure the same variable. For example, the approaches of Pinder and Harlos (11) and Tangirala and Ramanujam (12) agree on the intentional withholding of information, but they identify different “targets” from whom employees withhold information; while Pinder and Harlos (11) refer to “persons who are perceived to be capable of effecting change or redress” (p. 334), Tangirala and Ramanujam (12) refer to other members of the workgroup. Such a differentiation, however, has significant implications, as the second case applies to silence between members of the team regardless of their position while the first definition implies withholding information from persons higher in the hierarchy or with power to effect change. Thus, based on the definition, one runs the risk of excluding certain behaviors from the “silence spectrum”; the definition we use to measure and “diagnose” employee silence will shape the findings and these in turn, will shape interventions and policies. Congruently, voicing concerns can often take the form of venting or complaining about a colleague’s or superior’s behavior; formulating “behavioral prototypes” on what is desirable and what not in terms of voice behaviors could potentially discourage genuine expression and sharing of crucial information—like for example in a strong culture of conflict avoidance (134). Moreover, very few studies collected data on the positive forms of employee silence (e.g., prosocial silence), which runs the risk of suggesting that there are no positive aspects to silence in the workplace. Thus, there is a research gap regarding the positive reasons for employee silence. Congruently, there is a significant heterogeneity in the measures identified and there seems to be a narrow understanding of what are the “appropriate” ways of speaking up, which might result in excluding important ways of sharing information.

A significant amount of research only investigated speaking-up content related to patient safety and quality of care—with some measures being developed on the overlap of employee voice with stating safety concerns (88) or with error disclosure (51). This runs the risk of suggesting an overlap between speaking-up for patient safety and employee voice in healthcare. While speaking-up for patient safety is of crucial importance, it remains unclear what other forms and content might be relevant to the employee voice notion in healthcare, and whether other important issues not directly related to patient safety concerns are being hindered when the focus is solely on patient safety and/or error reporting. Congruently, the studies involving external observation of speaking-up behaviors were directly related to patient safety concerns, and almost all occurred within simulation or training settings and in a rather controlled environment. In terms of ecological validity, it is impossible to know whether the findings could be replicated in real work settings, due to the numerous limitations associated with observational studies employing trainings and/or simulations.

Similarly, our review has identified several measures of employee voice/silence and speaking up that consist of items related to the participants’ perceptions of whether speaking-up is valued or frowned upon in their team/organization; such measures provide us with important insights into the voice-related culture of healthcare teams and organizations. However, it is rather unclear how these measures are related to or differentiated from similar measures of organizational culture and work-related climate (6, 135), as well as the potential conceptual/theoretical overlap with other concepts like psychological safety and teamwork climate.

Differences in interpretation will lead to different ways of formulating and addressing the problem. This will become an even bigger problem when the justification of measuring voice or silence in a certain way is not clearly related to a specific definition. The fragmentation related to the silence and voice research does however not only emerge from differences in concept definitions, but also results from an unstructured framework related to the two concepts. The various antecedents (e.g., condition and situations, issues that trigger voice/silence response, motives) are all mixed in various ways. Motives for silence are for example sometimes understood as feelings or conditions and other times as situations, while also being classified as forms of silence and/or voice. As indicated by the studies that use the Van Dyne et al. (53) scale, the types of prosocial silence, acquiescent silence and defensive silence emerge from associated motives, while Yalcin and Baykal (40) measured silence motives as silence climate, silence based on fear, acquiescent silence and silence based on protecting the organization. Thus, motives can be both feelings (e.g., fear) and can indicate perceptions of organizational conditions as well as the type of context which triggers the silence behavior (e.g., a relational context in which one wants to protect the organization or to cooperate smoothly with colleagues). Overall, the concepts of voice and silence are understood in very heterogenous ways, and the ways in which voice and silence are framed are fragmented and bleary. This creates a vague starting point when trying to identify suitable interventions in order to create conditions that promote voice in healthcare organizations and might be a reason for why existing interventions seem to fail.

The vast majority of the reviewed research relies upon self-reported measures—with the exception of the seven studies involving external observation. This means that any attempt to map the silence/voice distinction—meaning whether what is observed is behavioral activation or behavioral inhibition (13)—is highly dependent on the employees’ subjective experiences and perceptions of whether and when engaging in voice behaviors is viewed positively or negatively by their team, supervisors and/or organization or their profession. Research in the field of social media usage has indicated significant discrepancies between self-reported data on time spent on social media versus objective data collected via monitoring the users’ devices (136). In simple terms, we cannot claim that what we know about employee silence and voice in healthcare via self-reported measures is what is really happening, but it is rather what employees remember, perceive, think and want to report when given the opportunity to participate in a study, as well as their perceptions of how “safe” speaking-up might be for their career and their team. And presumably also, this is limited by what they consider being included in “speaking up”; for example, some employees might think “speaking up” only means making an official report, or conversely, that reporting does not constitute speaking-up, but only verbal mentions do.

This brings us back to the importance of the context, the persons involved and the content of silenc

留言 (0)

沒有登入
gif