A Retrospective Study on Infant Bed-sharing in a Clinical Practice Located in an Urban Cluster

Abstract

Objectives

This retrospective study evaluates infant bed-sharing at a Missouri family practice with OB care.

Methods

After Institutional Review Board (IRB) approval, data were extracted from the first four well-child visits of 2374 infants between Sept. 2003 and Dec. 2011.

Results

Bed-sharing decreased after 2005 (25%, 39%, respectively, p = 0.000). For infants who bed-shared, the frequency of bed-sharing did not decline.

Conclusions

Nearly 20% of infants bed-share before the first visit; safe sleep prenatal education is warranted.

Introduction

Studies have confirmed bed-sharing as a known risk factor for sleep-related infant deaths.^{1, 2} In 2005 the American Academy of Pediatrics’ (AAP) Task Force on Sudden Infant Death Syndrome (SIDS) stated a new recommendation to reduce the incidence of sleep-related deaths: “infants not bed-share during sleep.”³ In 2011 AAP’s Task Force on SIDS reiterated their previous recommendation against bed-sharing.⁴ Additionally, the 2011 Task Force stated “a crib, bassinet, or portable crib/play yard that conforms to the safety standards of the Consumer Product Safety Commission and ASTM International…is recommended” for every sleep.⁴ These recommendations were supported by many other institutions, most notably the U.S. Consumer Product Safety Commission^{5, 6} and the National Institute of Child Health and Human Development (NICHD).^{7, 8}

The 2011 Task Force also recommended “Continue research and surveillance on the risk factors, causes, and pathophysiological mechanisms of SIDS and other sleep-related infant deaths.”⁴

Since the AAP’s 2005 new recommendation against bed-sharing, we know of only four prevalence studies on bed-sharing in infants, six months of age or less, which included data after 2005: Hauck et al. study⁹ (May 2005 to June 2007), Krouse et al. 2008 study¹⁰ (A. Krouse, Phd, written communique, August 2013), the first author’s previous study¹¹, Norton and Grellner (March 2002 to April 2008), and the ongoing National Infant Sleep Position Study¹² (NISP) 1993–2010.

In the Hauck study,⁹ of approximately 2,300 infants nationwide, bed-sharing rates were as followed; 1 month of age, 42.5%; 3 months, 34.0%; and at 6 months, 30.7%. This study did not evaluate whether a change in the rate of bed-sharing occurred with the new AAP’s recommendations. The Krouse et al. study of 2008 was a limited study of 94 new mothers in Pennsylvania; 47% of mothers reported bed-sharing at one month of age and 17% at three months of age. ¹⁰

Norton and Grellner (2010),¹¹ studying the infant population of a clinical practice in an urban area: 1 month of age, 18%; 3 months, 18%; and at 6 months, 14%. NISP (1993–2010) study found bed-sharing rates increasing for black and Hispanic infants throughout the study period. However among white infants rates of bed-sharing did not change.¹² In the Norton and Grellner (2010) study,¹¹ the bed-sharing rate decreased significantly after AAP’s recommendation against bed-sharing in 2005.

The objectives of this study were fourfold:

1. to continue surveillance of the rate of bed-sharing;

2. to identify the prevalence of bed-sharing and other unsafe sleeping practices in an infant population of a rural/urban cluster;

3. to examine the independent variables associated with bed-sharing; and

4. compare this current study with the first author’s previous study noted above.

The term “bed-sharing” referred to a parent(s) and/or other adults or children sleeping with an infant.³

Methods

Study Population and Design

This study utilized electronic medical records (EMR) from a private practice of family medicine with obstetrical care (OB) in Missouri. The practice was located in an urban cluster (populations between 2,500–49,999) and was surrounded by rural areas as determined by the U.S. Census^13,14 of 2000 and 2010. The nearest urban area was 40 miles away. After approval for the study from the participating medical institution’s Institutional Review Board was received, a document query of the EMR from the institution was completed. We obtained the de-identified EMRs of all infants who had presented to the designated family practice office for any of their initial four well-child visits from September 2003-December 2011.

Infants were excluded from the study for three reasons:

1. no sleep location was recorded at any of the well-child visits;

2. their first well-child visit occurred prior to the beginning of use of EMRs in the office (Sept. 2003); or

3. their birth date (month and year) precluded completion of the fourth well-child visit before the end of study’s time frame. Thus, infants born between September 2003 and April 2011 comprised the total cohort. Chart reviews were completed by hand of each well-child visit as marked in the EMR. Age at the visits was categorized as follows: newborn to one month (first well-child visit), two to three months (second well-child visit), four to five months (third well-child visit), and six to eight months (fourth well-child visit). None of the infants in the previous study were in the current study.

Measurement of Dependent and Independent Variables

During routine well-child visits, data were recorded into the EMR by the health care professionals working in the family practice office. The office used the EMR system called “Centricity” (S. Fair, written communication, May 2013). The EMR contained a section labeled “Sleeping” which prompted health care professionals to inquire about the infant’s sleep environment. For the first four well-child visits, a simple checklist with three options were provided: “crib,” “bassinet,” “w/parent(s).” Addition, space were provided to add notes on sleep environment. Multiple sleep environments could be indicated.

Demographic variables obtained from the EMR included number of well-child visits, gender, social-economic status (SES), birth weight, hospital and/or ER visits, NICU admission, home apnea monitor use, social environment, and nutrition (i.e. all variables evaluated in the first author’s previously cited study). In addition three more variables were examined: bassinet use, infants noted as having a chronic illness, and mode of delivery. These demographic variables were selected for evaluation as they might influence the caregivers in sleep location for their infants. Bassinet usage was examined in relation to sleep location when bassinet use was discontinued.

Low SES was designated to infants whose families received Medicaid, MO HealthNet (Missouri’s health care assistance program), or who had no health insurance. High SES was designated by families with private or military medical insurance. The EMR recorded only one insurance type for each infant. Low birth weight (LBW) was defined as a weight < 2500 grams. An infant’s social environment was designated as “poor” (PSE) if health care professionals noted parental drug use, domestic violence, and/or Department of Family Services (DFS) involvement. Health status issues were analyzed using five parameters:

1. medical intervention--a report of a visit to the Emergency Department and/or hospitalization;

2. NICU admission with use of a home apnea monitor;

3. NICU only (NICU admission without use of home apnea monitor);

4. home apnea monitor use without recorded NICU admission; and

5. chronic illness as identified by the clinician at the third and/or fourth well-child visits. Infant nutrition was dichotomized between ever breastfed and not ever breastfed. Mode of delivery was either vaginal delivery (with or without instrumentation) or C-section.

Statistical Analysis

These demographics were analyzed using multiple Pearson’s chi-squared tests. All analyses were evaluated using a significance cutoff of p< .05. Analyses were run through version 19 of SPSS statistical software package (SPSS, Inc., Chicago, Ill).

Results

Between September 2003 and December 2011, a total of 2,374 infants presented for their well-child visits (See Table 1). Out of these, 853 infants were present at all four well-child visits at the identified medical center. Whereas, the remaining 1,521 infants did not attend one or more well-child visits at the designated office (707 infants attended the first well-child visit, 1,091 attended the second, 842 attended the third, and 723 attended the fourth visit). Sleep location was not recorded in 1.5% of 6,775 visits (n=99). In the study, 66% had low SES (n=1,423) with WIC participation at 76%.

Table 1.

Demographics

	Total cohort		Complete		Incomplete
	N	%	N	%	N	%
Final cohort (number of infants)	2374		853		1521
Complete: present at all 4 well-child visits	853	35.9
Incomplete: not present at all 4 well-child visits	1521	64.1
WIC1	1598	76.2	611	76.6	987	75.9
Male	1246	52.5	437	51.2	809	53.2
Low birth weight (LBW) < 2,500 g 2	90	6.5	47	6.4	43	6.6
Went to ER or Hospital	134	5.6	57	6.7	77	5.1
NICU and apnea monitor	15	.7	10	1.2	5	.3
NICU only	76	3.2	47	5.5	29	1.9
Apnea monitor	9	.4	3	.4	6	.4
PSE, domestic abuse, drug use, DFS	32	1.6	6	.7	26	1.7
Chronic illness	72	3.0	28	3.3	44	2.9
Private or military insurance3	751	34.5	286	36.4	465	33.5
Public or no insurance3	1423	65.5	500	63.6	923	66.5
Nutrition (number of infants with data)	2371		853		1518
Ever breastfed4	1059	44.7	467	54.8	592	39.0
Breastfed for ≤ 1 month	340	14.3	147	17.2	193	12.7
Breastfed for 2–3 months	213	9.0	68	8.0	145	9.55
Breastfed for 4–5 months	149	6.3	54	6.3	95	6.3
Breastfed for ≥ 6 month	357	15.1	198	23.2	159	10.5
Vaginal Delivery5	840	64.0	456	64.7	384	63.3

¹No data were available on 276 infants

²No data were available on 991 infants

³No data were available on 200 infants

⁴No data were available on 3 infants

⁵No data were available on 1062 infants

Bed-sharing was reported at least once during the four well-child visits for 29% of the total cohort (n=687) (See Table 2). Most families reported only bed-shared at one visit. Bed-sharing was most common at the first and second well-child visits with a decline in overall bed-sharing at the third and fourth well-child visits (See Table 2). Eighteen percent of infants bed-shared before their first well-child visit.

Table 2.

Bed-sharing

	Total cohort		Complete group		Incomplete group
	n = 2374	%	n =853	%	n = 1521	%
Ever documented bed-sharing	687	28.9	269	31.5	418	27.5
Bed-shared once only	422		140	16.4	282	18.5
Bed-shared twice only	160		63	7.4	97	6.4
Bed-shared three times only	82		43	5.0	39	2.6
Bed-shared all four times	23		23	2.7

Figure 1 shows the proportion of bed-sharing over the eight years of the study. It is important to note here that a full year of data was not collected on the first (2003) and last (2011) years on the graph. The rate of bed-sharing was analyzed from three perspectives. Infants born after 2005 (the year AAP first published its guidelines advising against bed-sharing) showed significantly less bed-sharing than those born in or before 2005 (25%, 39%, respectively, p<0.000). When looking at bed-sharing (yes/no) across years (a Spearman’s Correlation), results are significant, negative, linear correlation for the total cohort (r = −.125, p <.001) which means that the trend is for less people, in general, to bed-share as the years progressed. However, in looking at bed-sharing (0, 1, 2, 3, 4 times) across years (again a Spearman’s Correlation), which results in non-significant findings for the total cohort (r = .002, p = .952) meaning that no trend existed across years for all of these points.

Figure 1.

Breastfeeding was reported by 45% of caregivers (n=1,059); 15% of the infants breastfed for at least 6 months (n=357). As shown in Table 3 infants who were ever breastfed were more likely to bed-share (p<0.001) at all four well-child visits (p=0.005, p<0.001, p<0.001, p= 0.005, respectively). Of the infants who completed all four well-child and breastfed for at least six months, they were more likely to bed-share at the third and fourth visits (p=0.001, p<0.001 respectively).

Table 3.

When bed-sharing occurred

	Total cohort			Complete group			Incomplete group
	n = 2374	# visits with data	%	n = 853	# visits with data	%	n = 1521	# visits with data	%
At 1st WC	274	1531	17.9	143	839	17.0	131	692	18.9
At 2nd WC	345	1929	17.9	141	845	16.7	204	1084	18.8
At 3rd WC	241	1672	14.4	101	840	12.0	140	832	16.8
At 4th WC	220	1544	14.2	102	837	12.2	118	707	16.7

Those families with low SES were more likely to ever bed-sharing than families with high SES (26% vs 31%, p= 0.008). Upon analyzing each visit, this was significant at the third well child visit only (p=.027).

Interestingly, families reporting NICU admissions only were less likely to bed-share overall (n=76, p = 0.005). The LBW infants tended to show significantly less bed-sharing (n=90, p = 0.001). There was no significant difference in the rate of breastfeeding between LBW infants and non-LBW infants (p= .9).

When looking at the months that bed-sharing occurred, there were significant differences between months (p = 0.007). Figure 2 depicts the bed-sharing trend; bed-sharing was significantly higher in December through May than in June, July, September and November.

Although information was gathered on gender, WIC enrollment, bassinet usage, delivery method, hospitalization and/or ER visits, and chronic illness, none of these factors significantly altered bed-sharing in this sample (See Table 3). Because of a small sample sizes and/or small cell sizes, PSE, NICU patients who also used home apnea monitors, and those with just apnea monitors use could not be analyzed.

Alternative, non-approved sleep environments occurred in fewer than 2% of infants (n=58). Car seats and playpens were the most common alternative sleep environment (n=32) followed by swings (n=8). These fifty-eight infants were no more likely to bed-share at other well-child visits than infants sleeping in approved locations (p = .559).

Using chi-square tests, rates of infants in the complete group who “never” bed-shared in our study were compared with the rates available from Missouri Pregnancy Risk Assessment Monitoring System¹⁵ (PRAMS) (2007: 79.2% vs 36.4%; 2009: 77.6% vs 42.5%; 2010: 75.0% vs 45.5%, respectively). Infants in our study were statistically significantly more likely to have “never” bed-shared compared with infants in the Missouri PRAM study. P-values were <.001 for each of the three years.

Discussion

The results contains both encouraging and discouraging news. Because it only takes one episode of bed-sharing to result in a fatality, it is significant that overall there has been a general trend toward more individuals never bed-sharing. However, for those who bed-share, the frequency of bed-sharing is not declining. We theorize that a trend of increased bed sharing may be occurring (See Figure 1). Data for 2010 and potentially 2011 (partial year of data collection) depict the influx of increased bed sharing. We speculate that this increase in recent years could be a result of many factors including:

1. caregivers’ preference superseding safe sleep message;

2. lack of safe cribs;

3. ineffective guidance from health care community;

4. caregivers lack of awareness; or

5. potentially, the conflicting messages caregivers receive from various media.¹⁶

The rates of bed-sharing in our study different significantly from Missouri’s PRAMS data. Previous studies have demonstrated different rates of bed-sharing based on race/ethnicity; bed-sharing rates are highest in black infants followed by Hispanic infants and least in white infants.^12,17 Race/ethnicity data for the study population were not available to us. In our study area the demographics¹⁸ in 2010 were black, .3%; Hispanic, 7.0%; white, 91.5% while Missouri’s demographics¹⁹ in 2010 were black, 11.6%; Hispanic, 3.6%; white, 82.8%. The lower rate of bed-sharing in our study could reflect the lower breast feeding initiation rates compared with Missouri’s PRAM study of rural women (54.8%, 66.1% respectively).²⁰ However the women in our study were less likely to discontinue breast feeding prior to 2 months than the PRAM study of rural women (31.2 %, 43.8%, respectively).²⁰

As in other studies, this study found bed-sharing is associated with infants ever having breastfed, ^{20, 21} breastfed for 6 months, and low social-economic status (SES). ^{21, 22, 23} In regard to independent variables associated with decreased bed-sharing (NICU only and LBW), the proportion of infants in both groups decreased after 2005.

In this study, seasonal distribution (December through May) was associated with higher bed-sharing than in June, July, September and November, whereas in the previous studies^{11, 24} no association had been discovered. The seasonal distribution found in this study mirrors the seasonality noted historically in SIDS with highest rates in December to March with lowest rates in June to September.²⁵ This variable warrants further investigation.

Bassinet use and bed-sharing were examined to determine if bassinet user indicated the lack of crib availability and thus subsequent increase rate of bed-sharing. The 1,121 bassinet users (47% of the total cohort) were not associated with an increased or decreased risk of bed-sharing. (See Table 4.)

Table 4.

Bassinet use

	Total cohort		Complete group		Incomplete group
	n = 2380	%	n =853	%	n = 1521	%
Ever documented bed-sharing	329	13.8	159	18.6	170	11.2
Bed-shared once only	203	8.5	85	10.0	118	7.8
Bed-shared twice only	85	3.6	45	5.3	40	2.6
Bed-sheared three times only	34	1.4	22	2.6	12	0.8
Bed-shared all four times	7	0.3	7	0.8

Recent studies have highlighted the need for NICU nurses to implement AAP’s safe sleep practices in the NICUs and to educate parents on SIDS risk-reduction strategies.^{26, 27} The decrease rate of bed-sharing of NICU infants in both studies is noteworthy. There was no significant difference in the rate of breastfeeding between NICU infants and non-NICU infants to explain the decrease rate of bed-sharing (p= .990).

An incidental finding was the perusal of the “Guidance” section of the EMR available to the healthcare professional. A healthcare provider could select from a list of over twenty potential topics that were discussed with a patient or write a topic in a comment section. “Co-sleeping” was a topic in the EMR list. In the NISP (1993–2010) study, caregivers that reported being advised by their physician against bed-sharing were less likely to bed-share.¹² EMRs designed to aid the physician in guidance on bed-sharing has the potential to decrease the incidence of bed-sharing.

One goal of this study was to replicate the previous study (a residency-based, family practice office with OB in an urban area) of Norton and Grellner (2010) for comparison to the current study (a private family practice with OB in an urban cluster). Overall, the current study replicated many of the previous findings. Both studies had similar reported rates of bed-sharing at least once during the four well-child visits (current study 29%; previous study 28%); a decrease in bed-sharing after 2005; the same three variables associated with bed-sharing: ever reported breastfeeding, breastfeeding for at least 6 months, and low SES. Similarly, NICU only infants and LBW infants had a decrease in reported incidence of bed-sharing. (See Table 5.)

Table 5.

Associations

Chi-square P-value
Associations with increased bed sharing
Ever reported breast feeding	0.000*
Breast fed at least 6 months (complete group)	0.005*
Public or no health insurance (Low SES)	0.008*
Bed-sharing as reported at each individual visit		≤ 1 month	2–3 months	4–5 months	6–8 months
Incomplete cohort (infants with less than 4 WC visits)		0.338	0.225	0.005*	0.012*
Ever reported breastfeeding		0.005*	0.000*	0.000*	0.005*
Breastfed at least 6 months (complete group)		0.243	0.053	0.001*	0.000*
Public or no health insurance (Low SES)		0.775	0.284	0.027*	0.643
Associate with decreased bed-sharing
Before vs. After 2005	0.000*
NICU	0.005*
Low birth weight	0.001*
Seasonal distribution	0.007*
Bed-sharing as reported at each individual visit		≤ 1 month	2–3 months	4–5 months	6–8 months
Low birth weight		0.001*	0.258	----	----
No association with bed-sharing
Gender	0.272
WIC	0.137
Method of delivery	0.211
Hospitalization or ER visit	0.527
Chronic Illness	0.759
Bassinet Use	0.618

Note. The symbols “---“ indicate a field that was unable to be analyzed due to small cell sizes.

^*denotes statistical significance

Limitations

All bed-sharing data relied on the knowledge and honesty of the caregiver accompanying the infant to the well-child visit. Sleep location data depended on the health care professional follow through with the EMR prompts, both for obtaining the data and accuracy of recording.

Secondly, in the previous study not all well-child visits need have occurred at the designated family practice office in an urban area; some well-child visits could have come from any one of numerous clinics in the health care system. While in the current study all well-child visits occurred at the designated family practice office in an urban cluster.

Birth weight data were unavailable for 42% of the infants. Because birth weight was usually recorded at the first well-child visit and only occasionally at subsequent visits. LBW would be under reported in 55% of the infants who were not at the first well-child visit. Similarly, NICU admissions could be under reported in this group. Lack of data on ethnicity, maternal age, educational background, infant birth order, smoking, and alcohol use of household members, sleep position, and place of residence were other limitations to the study.

Infant Bed-Sharing Recommendations

With nearly 20% of infants bed-sharing before the first well-child visit, providing patient education on safe beginning in the third trimester of pregnancy is warranted. If mothers are unprepared for a safe sleeping environment at the time of delivery, they will have difficulty arranging a safe sleep environment for the day of hospital discharge. Implementing education before delivery gives the family the time and information needed in order to prepare a safe sleep environment before the baby’s arrival home. All residency programs for physicians providing obstetrical care need to include instruction to the residents on the most current safe sleeping practices of infants.

We recommend that EMR include a prompt for the medical care professional to counsel caregivers on safe sleeping practices including the risk of bed-sharing. Evaluation of EMR designs might highlight possible improvements to aid the physician with patient education in this area. We also recommend that software programs be written for and used with EMR systems to facilitate data retrieval 1) for surveillance of risk factors associated with SIDS and other sleep-related infant deaths and 2) for evaluation of clinical and public health interventions aimed at reducing unsafe sleep environments.

Biography

Patricia J. Norton, MD, (above) is a graduate student, Master of Public Health, Missouri State University, Springfield, Mo., and Kathrene D. Valentine, BS, MS-II, is a candidate for a PhD in Psychology, emphasis in Quantitative Psychology, University of Missouri - Columbia.

Contact: patricia.j.norton@gmail.com