The Effect of Hospital Visitor Policies on Patients, Their Visitors, and Health Care Providers During the COVID-19 Pandemic: A Systematic Review.

Health care policymaking during the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic has questioned the precedent of restricting hospital visitors. We aimed to synthesize available data describing the resulting impact on patient, family/visitor, and health care provider well-being. We systematically reviewed articles from the World Health Organization COVID-19 Global Literature on Coronavirus Disease Database published between December 2019 through April 2021. Included studies focused on hospitalized patients and reported 1 or more prespecified main or secondary outcome (coronavirus disease 2019 [COVID-19] disease transmission, global well-being, mortality, morbidity, or health care resource utilization). Two authors independently extracted data into a standardized form with a third author resolving discrepancies. A total of 1153 abstracts were screened, and 26 final full-text articles were included. Ten studies were qualitative, with 7 cohort studies, and no randomized controlled trials. Critically ill patients were the most represented (12 out of 26 studies). Blanket hospital visitor policies were associated with failure to address the unique needs of patients, their visitors, and health care providers in various clinical environments. Overall, a patient-centered, thoughtful, and nuanced approach to hospital visitor policies is likely to benefit all stakeholders while minimizing potential harms.

Restrictive hospital visitor policies have differential effects on various populations, notably critically ill patients, clinicians, and family or support persons.

There may be challenges with effective and equitable implementation of video conferencing.

The effect of altering hospital visitor policies on coronavirus disease 2019 (COVID-19) transmission is poorly studied.

It is prudent to pursue a patient-centered, thoughtful, and nuanced approach to hospital visitor policies accounting for clinical judgment as opposed to blanket rulings produced by administrators.

Alt-text: Unlabelled box

SEE RELATED ARTICLES, p XXX.

Alt-text: Unlabelled box

Introduction

In western tradition, only early in the 20th century did hospitals begin to allow visitors for paying patients. Fifty years later, after the establishment of newborn intensive care units (NICUs), visitor policies appeared more familiar to those of modern day with limited visiting hours for all patients. Additional restrictions for infection control is an established practice during respiratory syncytial virus and influenza seasons.3, 4, 5 Although not novel, the efficacy and guidance for visitor restrictions remains inconsistent, especially for the coronavirus disease 2019 (COVID-19) pandemic. The Centers for Disease Control and Prevention (CDC) recommended limiting visitors to inpatient facilities “to only those essential for the patient's physical or emotional well-being and care.” It is reasonable to suspect that practical application of this statement may vary across institutions and practices. Lack of clarity leaves the potential for inequities in implementation and raises ethical questions. , Restriction of visitors is also discussed as a source of moral distress for health care providers who may not agree with hospital policies. , Yet, a Cochrane rapid qualitative evidence synthesis review of barriers to health care providers’ adherence to infection control measures found an opposing effect. Health care workers experienced strain from being responsible for visitors’ adherence to infection control measures. The significance of the transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) from visitors in hospitals, however, is also questionable. Therefore, we sought to critically appraise the evidence relating hospital visitor restrictions and COVID-19 transmission. We aimed to determine the impact of visitor restrictions on the well-being of hospitalized patients, their visitors, and health care providers during the COVID-19 pandemic.

Methods

Further details are available in the PROSPERO protocol associated with this study (CRD42021248603) that was developed in accordance with the preferred reporting items for systematic reviews and meta-analyses (PRISMA) checklist.

Data Sources and Searches

We searched the World Health Organization (WHO) COVID-19 Global Literature on Coronavirus Disease Database, which includes more than 30 databases, at https://www.who.int/publications/m/item/quick-search-guide-who-covid-19-database, using the following search strategy: (ti:(visit*)) OR (ab:(visit* AND (restrict*))) OR (mj:(MH:"Visitors to Patients")).

Study Selection

We included quantitative and qualitative studies as well as conference abstracts from December 2019 to April 2021. Studies must have focused on hospitalized patients, their families and visitors, or health care providers of all ages in the setting of the COVID-19 pandemic. “Hospital” was defined as a public or academic institution in which a patient was admitted for inpatient medical care. Studies were in English and reported at least 1 of the prespecified main or secondary outcomes (COVID-19 disease transmission, global well-being, mortality, morbidity, or health care resource utilization).

Reviews of existing literature, editorials, and expert opinions were excluded. Studies that did not fit into the conceptual framework of this systematic review or focused on a population other than hospitalized patients were also excluded. Long-term care and skilled nursing facilities were excluded because they are not considered hospitals but places of permanent residence.

Using a systematic review software, Rayyan, 2 independent reviewers screened abstracts based on the predefined criteria. Discrepancies were discussed with a third reviewer until a consensus was reached. This same process was repeated with full articles.

Data Extraction and Quality Assessment

For each included study, 2 reviewers extracted data independently using a standardized data extraction form (Supplementary Table 1, available online). This process occurred without blinding of study authors, institutions, journals, or results. Discrepancies were resolved by discussion with a third reviewer and the research team, as necessary.

Supplementary Table 1

Data Extraction Form

Criteria	Data entry
Study characteristics and data quality
General	Article ID #, First author, Journal, Year of publication
Geographic location	0=USA (please list state[s]), 1=Other country (please list)
Setting	0=General ward, 1=Stepdown, 2=ICU
Level of Care	0=Regular visit, 1=End-of-life care, 2=Other special circumstance
Type	0=Clinical trial/interventional, 1=Observational
Study design	General design, Inclusion criteria, Exclusion criteria
Study quality and risk of bias	GRADE: 0=not needed, 1=very low, 2=low, 3=moderate, 4=high
	ROBINS-I (if applicable), ROB 2 (if applicable), CASP (if applicable)
Studies of patients
Sample size
Patient population	0=Adults (mean age >18), 1=Geriatrics (mean age >65), 2=Neonates (mean age <4 weeks), 4=Pediatrics (mean age >4 weeks, <18 years)
	Percent male
	Special population
	COVID-19 status: 0=Diagnosed, 1=Symptomatic, 2=Suspected, 3=No COVID
Visitation policy	Type and description: 0=Unrestricted, 1=Some restrictions, 2=No visitors
Outcome measures (method and result)	Primary: Global well-being, Quality of life, Satisfaction survey, Moral distress
	Secondary: Morbidity, Mortality, Health care resource utilization
Studies of health care providers
Sample size
Subjects	0=Physician, 1=Student, 2=Nurse, 3=Ancillary staff
	Percent male
Visitation policy	Type and description: 0=Unrestricted,1=Some restrictions, 2=No visitors
Outcome measures (method and result)	Primary: Global well-being, Quality of life, Satisfaction survey, Moral distress
	Secondary: Morbidity, Mortality, Healthcare resource utilization
Studies of family members
Sample size
Patient population	0=Adults (mean age >18), 1=Geriatrics (mean age >65), 2=Neonates (mean age <4 weeks), 4=Pediatrics (mean age >4 weeks, <18 years)
	Percent male
	Special population
	COVID-19 status: 0=Diagnosed, 1=Symptomatic, 2=Suspected, 3=No COVID
Visitation policy	Type and description: 0=Unrestricted,1=Some restrictions, 2=No visitors
Outcome measures (method and result)	Primary: Global well-being, Quality of life, Satisfaction survey, Moral distress
	Secondary: Morbidity, Mortality, Health care resource utilization

CASP = Critical Appraisal Skills Programme; COVID-19 = coronavirus disease 2019; ICU = intensive care unit; ROBINS-I = Risk Of Bias In Nonrandomized Studies of Interventions.

We evaluated the risk of bias using the Cochrane Collaboration's tool Risk Of Bias In Non randomized Studies of Interventions (ROBINS-I), Critical Appraisal Skills Programme (CASP) appraisal tool for qualitative studies, and Oxford Centre for Evidence-based Medicine quality scheme.15, 16, 17 Two authors ranked each study's risk of bias separately. Disagreements were resolved by discussion with a third reviewer.

Data Synthesis and Analysis

Given the heterogeneity in methodologies across the literature, we comprehensively tabulated study characteristics, permitting critique of design and enumeration of potential biases. Two authors (ANI and JOA) independently made this determination, with input from a third author in the event of a discrepancy (WS). A narrative synthesis was performed including study design, patient population, hospital visitor policy changes, and reported outcomes. Mean COVID-19 transmission rate was the primary outcome. Thorough documentation was prioritized given the paucity of high-quality studies on this emerging topic and with predominance of descriptive outcomes. Studies were grouped by population (eg, neonatal, pediatric, or adult patients), clinical setting (eg, intensive care unit, general floor unit, end-of-life care vs general care, etc.), and visitation policy. Data were handled by groups as opposed to individual participant data.

Results

Description of Included Studies

We identified 1153 abstracts, which were read in full. A total of 78 articles met criteria for screening of the study report. Of these, 26 studies met prespecified criteria for inclusion (Figure 1 ). Half of the studies were published in 2020, the first year of the pandemic and half in 2021. Most studies took place in North America (46%), specifically the United States, followed by Europe (42%), including the UK, France, and Italy. Patients were the population of interest in 37% of included studies; family, support persons, or visitors were the focus of 33%; 29% assessed the impact of visitor restrictions on health care providers; and the remaining 11% addressed multiple populations. As may be expected with severe COVID-19 infection, a plurality of studies took place in intensive care units (ICUs) (Figure 2 A). Finally, the majority of the included studies were qualitative or survey-based; only 7 of 26 were cohort studies (Figure 2B).

Figure 1

Systematic review protocol workflow. Flowchart is adapted from the PRISMA flow diagram for included studies. PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Figure 2

Characterization of included studies. (A) Clinical context of included studies broken down by clinical setting and level of care/acuity with most studies taking place in the ICU. (B) Graph tabulating the number of each study methodology. For all, N = 26. ICU = intensive care unit.

Inpatient Wards

We found 5 of 26 included studies focused on the inpatient ward setting, and a common theme among them was attention to vulnerable populations with potentially impaired decision-making capacity (Table 1 ). Visitors or support persons may serve as patient advocates and aid in decision-making if legally authorized. One common cause of impaired capacity in the inpatient setting is delirium. Some suggest that the presence of hospital visitors may protect against delirium development, but this notion was challenged in recent studies. , A retrospective cohort study of adults who underwent emergency hospital admissions in Japan (N = 6264) found greater odds for developing delirium after pandemic visitor restrictions as compared with unrestricted visitor policies (adjusted odds ratio [AOR] for delirium incidence 3.79, 95% confidence interval [CI], 2.70-5.31). Similarly, patients in inpatient psychiatric facilities also represent a vulnerable population because of their often limited autonomy and potentially diminished capacity and were negatively impacted by visitor restrictions. In 1 study of 24 patients in Norway, 80% reported difficulty with not having loved ones present and expressed a need for visitors, 64% believed staff would prevent them from becoming infected with COVID-19, and 88% overall felt safe at the ward. On the contrary, of 140 staff members surveyed at the same facility, 57% feared bringing home COVID-19 from work and 69% were satisfied with measures to mitigate risk of infection (including visitor restrictions). Two additional potentially vulnerable populations include pediatrics and peripartum patients, which are discussed in the Supplemental Materials, available online.

Table 1

Summary of Included Studies Focused on the Inpatient Ward Setting*

First Author	Location	Design	Sample size	Participants	Visitor policy
Dieset I23	Norway	Cross-sectional survey	24	Psychiatric inpatient (female: 69%)	No visitors
			140	Staff in an acute psychiatric inpatient unit (nurses: 67%, physicians or psychologists: 26%, other staff: 7%)
Kandori K21	Japan	Retrospective cohort	6264	Adult emergency admissions inpatients (median age: 74, female: 47%%)	No visitors
Onwuteaka-Philipsen BD24	Netherlands	Cross-sectional survey	241	Health care professionals (female: 49%) regarding patients who died	Some restriction
Feder S26	United States	Cross-sectional survey	328	Next of kin of veterans who died in an inpatient unit (mean age: 76, female: 7%)	No visitors
Maaskant JM25	Netherlands	Cross-sectional survey	9	Bedside nurses of inpatient COVID-19 hospital wards (median age: 32, female: 89%)	Some restriction

COVID-19 = coronavirus disease 2019.

The “No visitors” designation was reserved for studies that explicitly stated no visitors were allowed under any circumstances, per hospital policy. Gender and mean or median age reported as available.

Several studies also concerned end-of-life care. A cross-sectional survey of health care providers regarding deceased patients in the Netherlands found sufficient emotional support was less common for patients for whom there had been restricted (AOR 0.46, 95% CI 0.29-0.75) or no visitors (AOR 0.23, 95% CI 0.12-0.45). Spiritual care was also less sufficient for patients who were not allowed visitors in the last 2 days of life (AOR 0.21, 95% CI 0.10-0.42). A small focus group (N = 9) of nurses discussed how the need for family involvement increases during end-of-life care, which can strain personnel who are tasked with communicating with families remotely. Similarly, a cross-sectional survey of 328 next of kin of veterans who died in an inpatient unit found themes of “anguish and despair” from not being allowed to see patients.

Overall, the COVID-19 pandemic and visitor restrictions were associated with negative emotions among most inpatients and their families in the studies reviewed, especially in the context of end-of-life care. There was evidence of moral distress for health care providers caring for dying patients; however, some responded positively to visitor restrictions from the perspective of limiting their occupational exposure to the COVID-19 virus.

Adult ICUs

End-of-life care and vulnerable population discussions are especially pertinent to the ICU setting (Table 2 ). A large cross-sectional survey (N = 1058) of health care providers in France reported that 31.5% regretted restricted visitation policies that were associated with symptoms of anxiety (odds ratio 1.39; 95% CI 1.03-1.86) and depression (odds ratio 1.49; 95% CI, 1.09-2.04). Clinicians in the UK also expressed dissent with restricting visitors and raised concerns about the detrimental effect on patients, their families, and staff.

Table 2

Summary of Included Studies Focused on the ICU Setting*

First author	Location	Design	Sample size	Participants	Visitor policy
Azoulay E27	France	Cross-sectional survey	1058	Frontline health care providers (median age: 33, female: 71%)	Some restriction
Azad TD29	Maryland, United States	Retrospective cohort	940	Adult decedents	Some restriction
Piscitello GM30	Illinois, United States	Retrospective cohort	61	Patients lacking medical decision-making capacity	Some restriction
Cattelan J31	France	Prospective cohort	88	Reference persons of patients referred to ICU for COVID-19 (female: 65%, first-degree relatives of the patient: 92%)	No visitors
Chen C32	New York, United States	Cross-sectional survey	10	Family members of adult patients with COVID-19 in the ICU	No visitors
Creutzfeldt CJ33	Washington, United States	Cross-sectional survey	19	Family members of patients with severe acute brain injury (female: 59%)	No visitors
Kennedy NR34	Pennsylvania, United States	Qualitative interview	21	Family members cardiothoracic and neurologic ICU patients (mean age: 56, female: 76%)	Some restriction
			14	Treating clinicians of cardiothoracic and neurologic ICU patients (female: 36%)
Sasangohar F35	Texas, United States	Cross-sectional survey	230	Family members of intensive care unit patients	No visitors
Jeyabraba S36	UK	Retrospective survey	24	Families affected by the visitor restrictions during the coronavirus period.	No visitors
Rose L37	UK	Cross-sectional survey	117 ICUs	217 UK hospitals with at least 1 ICU	Mixed
Azoulay E27	France	Cross-sectional survey	1058	Frontline health care providers (median age: 33, female: 71%)	Some restriction
Boulton AJ28	UK	Cross-sectional survey	134	ICUs with patients with COVID-19	Some restriction

COVID-19 = coronavirus disease 2019; ICU = intensive care unit.

The “No visitors” designation was reserved for studies that explicitly stated no visitors were allowed under any circumstances, per hospital policy. Gender and mean or median age reported as available.

Video conferencing is becoming a common means of communicating with families remotely; however, it is uncertain whether this can fully substitute for physical presence. A retrospective cohort of 940 patients with adult descendant subjects found that patient admission after restrictive visitor policy implementation was associated with a significantly longer time to first do not resuscitate, do not intubate, or comfort care order (adjusted hazard ratio, 2.2; 95% CI, 1.6-3.1). Another retrospective cohort study of patients lacking medical decision-making capacity found more frequent changes in patient goals of care for in-person meetings than by video (36% vs 11%, P = .0006), implying a potential differential effect of communication modality on outcome. A prospective cohort of 88 families of ICU patients in France assessed the impact of remote communication on reference persons (RPs)—family members as designated points of contact for communication from health care providers. All RPs surveyed described a specific type of “responsibility” with being an RP in a remote-only context, leading RPs to restrict communication to extended family members (67%). Thirty percent of RPs related the situation to a prior traumatic experience. Among all the studies assessed here, restriction of visitors in the ICU setting was associated with distress among patients and their health care providers. A prominent challenge in this environment was navigating goals of care discussions and it seems that mode of communication may influence the decision-making outcome.

Patients in the Postoperative Period

Two studies covered the postoperative patient population (Table 3 ). A retrospective cohort study of 117 postoperative patients who were not allowed visitors found that those in the no-visitor cohort were less likely to report complete satisfaction with the hospital experience (80.7% vs 66%, P = .044), timely receipt of medications (84.5% vs 69%, P = .048), and assistance getting out of bed (70.7% vs 51.7%, P = .036). Qualitative analysis of patient responses highlighted the consistent psychosocial support provided by visitors after surgery (84.5%), and patients in the no-visitor cohort reported social isolation due to lack of psychosocial support (50.8%). A similar study of 541 postoperative patients in Italy, however, found a unique benefit of pandemic precautions for postoperative patients. The measures to reduce the severe acute respiratory coronavirus 2 spread (ie, surgical mask wearing and no visitors allowed) were independently associated with the reduction of all surgery site infections (AOR: 0.316, 95% CI: 0.103-0.970). It seems that psychosocial stress results from limiting visitor access may be somewhat offset by reducing infection risk in the postoperative setting. Interestingly, the transmission of COVID-19 infection was still not assessed.

Table 3

Summary of Included Studies Focused on the Postoperative Setting*

First Author	Location	Design	Sample size	Participants	Visitor policy
Zeh RD38	Ohio, United States	Retrospective cohort	117	Postoperative surgery patients (mean age: 57.5, female: 56.4%)	No visitors
Losurdo P39	Italy	Retrospective cohort	541	Surgical patients (mean age: 62, female: 59.8%)	No visitors

The “No visitors” designation was reserved for studies that explicitly stated no visitors were allowed under any circumstances, per hospital policy. Demographic information reported as available.

Video Conferencing

In lieu of in-person visits, video calls are increasingly prevalent in hospitals. A retrospective survey of families (N = 24) in the UK who received video calls were more likely to be satisfied with the frequency (83%) and quality (83%) of communication. All reported that video calls helped them to stay connected with the patient and health care team. Although another valence-based and manual sentiment analysis of family members of ICU patients found mostly positive responses to video visits, 44% of respondents (25 out of 57) found it challenging to communicate with patients due to being either intubated or undergoing procedures, a pertinent consideration for the critically ill patient population. A small (N = 10) cross-sectional qualitative survey of family members of ventilated patients highlighted this challenge. This group struggled to feel informed, had difficulty understanding information, and found video calls unhelpful. Commonly reported benefits for those patients who could participate in virtual visits across 117 ICUs in the UK included reducing patient psychological distress (78%) and reorientation of delirious patients (47%).

Qualitative interviews with ICU physicians yielded different results. This study found that telehealth increased clinician workload, technical difficulties limited effective communication, and clinicians struggled to engage family members because of discomfort with technology use and less apparent social cues. Clinicians also were concerned about ability to convey empathy remotely. Family member participants, however, felt empathy was relayed successfully via phone and video. In this same study, respondents viewed phone and video communication as somewhat effective but inferior to in-person communication. Common barriers to virtual visiting included challenges associated with family member ability to use videoconferencing technology or having access to a device. Communication strategies suggested by families and clinicians for productive remote interactions include identifying a family point person to receive updates, frequently assessing family understanding, positioning the camera so that family can see the patient, and allowing time for the family and patient to interact without clinician presence. Interestingly, this model is in contrast to the study in France that reported high levels of stress among RPs who were to field provider calls and updates. Overall, judicious use of video conferencing may be beneficial in some settings. Optimizing communication strategies is important from the provider standpoint. Technical issues and access to technology may limit effective implementation of video conferencing.

Study Quality

The authors found the included studies were of limited quality (Table 4 ). Qualitative studies, of which there were several (Figure 2B), were evaluated using the Critical Appraisal Skills Program appraisal tool. Here, the authors found inconsistent use of validated formal assessment measures for coding of subject interviews. Globally, this contributed to substantial heterogeneity, limiting the ability to synthesize data. Furthermore, the nature of the secondary research questions has a tendency toward qualitative studies which, by nature, precludes most quantitative analyses. Given the uncertain potential risks and benefits of allowing hospital visitors, it might be argued that a randomized controlled trial is justified in this situation to provide further clarity, as was done for a study assessing impact of visitor presence on delirium prevalence in patients in the ICU.

Table 4

Evaluation of Study Quality*

First Author	Location	Design	Modified OCEM Rating	ROBINS-I Score or CASP Rank
Ashini A40	Libya	Cross-sectional survey	4	10 (CASP)
Azad TD29	United States	Retrospective cohort	3	3 (ROBINS-I)
Azoulay E27	France	Cross-sectional survey	4	3 (CASP)
Bembich S41	Italy	Cross-sectional survey	4	6 (CASP)
Boulton AJ 28	UK	Cross-sectional survey	4	9 (CASP)
Bradfield Z42	Australia	Cross-sectional survey	4	3 (CASP)
Cattelan J31	France	Prospective cohort	3	3 (CASP)
Chen C32	United States	Cross-sectional survey	4	2 (CASP)
Creutzfeldt CJ33	United States	Cross-sectional survey	4	9 (CASP)
Cullen S43	Ireland	Cross-sectional survey	4	6 (CASP)
Dieset I23	Norway	Cross-sectional survey	4	7 (CASP)
Feder S26	United States	Cross-sectional survey	4	3 (CASP)
Jeyabraba S36	UK	Retrospective survey	4	6 (CASP)
Kandori K21	Japan	Retrospective cohort	3	1 (CASP)
Kennedy NR34	United States	Qualitative interview	4	1 (CASP)
Losurdo P39	Italy	Retrospective cohort	3	2 (ROBINS-I)
Maaskant JM25	Netherlands	Cross-sectional survey	4	1(CASP)
Mayopoulos G44	United States	Cross-sectional survey	4	3 (CASP)
Muniraman H45	UK	Cross-sectional survey	4	3 (CASP)
Onwuteaka-Philipsen BD24	Netherlands	Cross-sectional survey	4	6 (CASP)
Piscitello GM30	United States	Retrospective cohort	3	3 (ROBINS-I)
Romano-Keeler J46	United States	Prospective cohort	3	8 (CASP)
Rose L37	UK	Cross-sectional survey	4	4 (CASP)
Sasangohar F35	United States	Cross-sectional survey	4	5 (CASP)
Wiener L47	United States	Cross-sectional survey	4	1 (CASP)
Zeh RD38	United States	Retrospective cohort	3	2 (ROBINS-I)

CASP = Critical Appraisal Skills Programme; OCEM = Oxford Centre for Evidence-based Medicine; ROBINS-I = Risk Of Bias In Nonrandomized Studies of Interventions.

Ratings based on Modified OCEM, ROBINS-I, and CASP scoring systems. For ROBINS-I, 0 = not needed, 1 = very low, 2 = low, 3 = moderate, 4 = high assessment of study quality.

Discussion

Risk-benefit assessment is the cornerstone of medical and policy decision-making. As our systematic review has shown, there are inadequate studies of sufficient quality to determine whether restrictive policies have been effective in potentially limiting the spread of COVID-19. Although it might be reasonable to speculate that these policies slow spread based on mechanistic understanding of disease, this must be weighed against the potential harms of limiting hospital visitors. Our study highlights the complexity associated with the numerous factors and stakeholders impacted by hospital visitor policies. A review of literature surrounding hospital visitor policies after the 2003 severe acute respiratory syndrome (SARS) outbreak concluded that there should be a tailored approach to visitor policies depending on the clinical scenario. For example, reasonable exceptions include allowing visitors for the labor and delivery unit, adults with disabilities, children, and dying individuals. Similarly, recommendations from 7 interdisciplinary German palliative care societies argue for making palliative care and dying patients exempt from hospital visitor bans. Specifically, the authors advocate for a patient-centered guidance of visitor policies. Together, this argues for a nuanced approach to hospital visitor policymaking, taking into account clinical setting, patient population, visitor use of personal protective equipment, screening measures, community prevalence, and immunization status. Exceptional circumstances should be included, and policies should be clearly articulated with transparent communication to the public. ,

The CDC now provides visitor recommendations for a variety of scenarios (eg, vaccinated versus unvaccinated status, symptomatic vs asymptomatic, and specific facilities, such as dialysis centers); however, end-of-life care is not addressed. The CDC also recommends using alternative methods of staying connected with patients, such as video calls. Although our findings suggest that increasing availability of video conferencing may be beneficial in certain situations, consideration should be given to how this practice may be practically and equitably applied. Exclusive reliance on technology may differentially impact those at an economic or resource disadvantage. The origin of this potential disparity harkens back to the infancy of hospital visitor policies when only paying patients were allowed to have visitors.

Finally, there is a paucity of reporting for COVID-19 transmission in the context of altering hospital visitor policies and the body of literature is mostly of limited quality. Further retrospective, but importantly prospective or randomized studies, are needed to clarify the risks and benefits associated with limiting hospital visitors. In the meantime, it is prudent to take a patient-centered and thoughtful approach to hospital visitor policies with freedom given to practicing physicians to grant exemptions as opposed to blanket rulings produced by administrators.

Strengths and Limitations

Strengths of our study include a systematic investigation of the available literature at a period of high policy turnover during the COVID-19 pandemic. Study parameters were prespecified in the protocol to reduce the risk of bias in accordance with accepted systematic review guidelines. Multiple authors independently completed each step to reduce noise among the otherwise heterogenous data. The most notable limitation was the inability to assess our primary outcome: the impact of visitor policies on COVID-19 transmission. Only 1 included study reported COVID-19 transmission, which was in the context of transmission from mother to newborn. Therefore, it is difficult to comment on whether restriction of hospital visitors significantly reduced the spread of COVID-19 infection. Second, our analysis was limited by the lack of reporting of COVID-19 status in study participants and pertinent details about the visitor policies in place. This may be due to fluid policies in the face of a rapidly evolving pandemic; however, a few investigators used this as an advantage to perform cohort studies. Notably, these retrospective cohort studies compared groups of patients pre- and postimplementation of visitor policy. , , , Potential downsides to these retrospective studies include biases related to the selection of subjects, recall bias, and confounding factors given lack of randomization. Additionally, 2 prospective cohort studies had relatively small sample sizes and only 1 assessed COVID-19 transmission. , Finally, the ROBINS-I tool was used, and all authors agreed that these studies were of low quality. Together, the data reflect an early, developing literature exploring the efficacy of policy changes in the face of a challenging pandemic.

Supplementary Table 2

Summary of Included Studies Focused on the NICU and Peripartum Settings*

First Author	Location	Design	Sample size	Participants	Visitor policy
NICU and PICU
Romano-Keeler J46	Illinois, United States	Prospective cohort	21	Newborns in the NICU, born to mothers who were COVID-19 positive	No visitors
Bembich S41	Italy	Cross-sectional survey	10	Parents of newborns in the NICU (mothers: 90%)	Some restriction
Muniraman H45	United States, UK	Cross-sectional survey	224	Parents of newborns in the NICU (mean age: 32, mothers: 70%)	Some restriction
Ashini A40	Libya	Cross-sectional survey	41	Parents of neonate(s) in the NICU (mean age: 32, mothers: 75.5%)	Some restriction
Wiener L47†	United States	Cross-sectional survey	207	Pediatric palliative care team members from 80 cities within 39 states	Mixed
Peripartum/Labor and Delivery
Cullen S43	Ireland	Cross-sectional survey	422	Antenatal or postpartum patients	Some restriction
Mayopoulos G44	Massachusetts, United States	Cross-sectional survey	136	Postpartum women (mean age: 32, 68 COVID-19 positive, suspected, or confirmed and 68 COVID-19 negative)	Some restriction
Bradfield Z42	Australia	Cross-sectional survey	620	Midwives (age ≥18, female: 98.5%)	Mixed

COVID-19 = coronavirus disease 2019; ICU = intensive care unit; NICU = neonatal intensive care unit; PICU = pediatric intensive care unit.

The “No visitors” designation was reserved for studies that explicitly stated no visitors were allowed under any circumstances, per hospital policy. Demographic information reported as available.

Study included both inpatient ward and ICU settings.