Discharge from an intensive care nursery should be dependent on the infant's clinical and social condition and independent of the day of the week.
Objective. To evaluate admission and discharge dates of 5272 neonates cared for in 5 major metropolitan regions in the United States and managed by a national disease management company for the distribution of the day of the week.
Study Design. All infants discharged to home between July 1, 1996 and September 30, 1998 are included. Data are represented as a percentage of total discharges or admissions for each weekday assignment. Using the normal approximation to the multinomial distribution, we tested for proportional differences on each weekday.
Results. The data demonstrate that the timing of nursery discharge has an uneven distribution across the days of the week, with weekend (Saturday and Sunday) discharge rates that are significantly lower than weekday discharge rates. This uneven distribution exists in both the term and preterm subgroups as well. There is also an uneven distribution of births among the days of the week, with a pattern that reveals fewer weekend births than weekday births in the entire population studied, as well as in both the term and preterm subgroups. Normalizing these weekend discharges to the previous weekday could generate potential saving of $1 569 405 in charges for the total population and 627 days of hospitalization. The average length of stay of infants discharged on Mondays is longer than for those infants discharged on Saturday or Sunday.
Conclusions. We speculate that changes in discharge planning could decrease the variation in day of discharge, shorten length of hospitalization, and potentially reduce cost.
The timing of discharge from an intensive care nursery (ICN) requires attainment of a number of factors so that neonates can smoothly transfer to life at home for the first time. Such factors include the resolution of medical problems, an adequate home environment, the scheduling of home care needs, and parental teaching and preparation.1,2 Recent changes in health care reimbursement have placed added importance on earlier discharge of these infants. In addition, optimal timing of discharge should be based on infant criteria and family convenience, rather than staffing or hospital issues. We hypothesized that infants admitted to an ICN are delivered in a random manner with respect to the day of the week. As a result, the appropriate timing of discharge, therefore, should be independent of the day of the week. If hospital discharges were skewed to more convenient times, then this finding would represent a factor in discharge unrelated to infant or familial factors. To evaluate this we assessed infant ICN admission and discharge timing practices in different ICN settings.
A group of 5272 neonates cared for in ICNs in 5 major metropolitan areas were studied. These infants were cared for in 223 different hospitals and grouped by geographic region for purposes of data analysis. Infants were managed by a national disease management (DM) company between July 1, 1996 and September 30, 1998. Former neonatal nurses who functioned as case managers entered clinical and demographic information on infants into an administrative database for the DM company. Dates of discharge to home were assigned the corresponding weekday and these values were compared. Data were represented as a percentage of total discharges for each weekday. Infants were grouped into term and preterm populations. Term gestation was defined as ≥37 weeks' gestation at delivery by best obstetrical estimate or Ballard score if the former was unavailable. Using the normal approximation to the multinomial distribution, we tested for proportional differences in the discharges on each day of the week assigned.3 Data were validated internally by the DM company for accuracy with the patients' hospital record by random chart review by an independent review of 1% of patient records for data accuracy. The research team also used a number of algorithms to test for data consistency. These included such tests as the likelihood that a patient's zip code, hospital, and region would correspond. Significant differences were determined by a P value <.05 by using the normal approximation to the multinomial distribution.
To determine potential cost savings of an even distribution of discharge percentages by day of the week, the discharges were normalized to the mean percentage of discharge days. Weekend discharges were assumed to have occurred on the previous weekday, and the difference in charges for an inpatient day is a speculated saving. The charge data were obtained by Gibson et al4 as actual charges for patients involved in an accelerated discharge pilot study, as this study was performed in a similar population of patients. The charge data were sampled from a number of different institutions and averaged. Charges are estimated to be 40% greater than were costs during this portion of the hospitalization. Although it is most appropriate to use cost data for comparison, we did not have access to charge data for multiple institutions to demonstrate averages in the manuscript.
The average length of stay (ALOS) of these infants was calculated for each day of the week of discharge. This was done for the entire population, as well as the term and preterm subgroups. The ALOS for each weekday was compared using multiple comparison tests, and significant differences were determined by a P value <.05.
Of the 5272 infants studied, 2168 were term gestation and 3104 were preterm patients. The day of birth distribution for all infants is demonstrated in Fig 1. As shown, weekday birthdays are more common than are weekend day births (P < .001), represented by the light bars. In addition, there is a trend toward fewer births occurring on Mondays than on the other days of the week. An evaluation of all study infants reveals a discharge distribution as seen in the darker bars in Fig 1, as well. This reveals fewer discharges on weekend days than on weekdays (P < .001).
The days of the week term infants are born shows a distribution pattern depicted in Fig 2. Here the lighter bars demonstrate a pattern of fewer births on weekend days (P < .001). The day of discharge of the term population group is also depicted in this figure by the darker bars. This reveals an unequal distribution of discharges among all days (P < .001), with weekend day discharge occurring significantly less frequently than weekday discharge (P< .001). The most frequent discharge days overall are Tuesday, Wednesday, and Friday.
For the preterm infants, weekday deliveries are more common than weekend days, (P < .001); there is little difference among the weekday birth rates, as shown by the lighter bars. Both this and the distribution pattern of discharge days, the darker bars, are displayed in Fig 3. There is an unequal distribution of discharge days, with weekend day discharges occurring less commonly than weekday discharges (P < .001). The most common weekdays to be discharged are Tuesday, Wednesday, and Friday (P < .001).
In each of 5 major metropolitan regions, weekend day discharges are less common than weekday discharges for all birth weights (P < .05). However, the distribution of discharge percentages among weekdays differs in each region, as depicted in Table 1.
Potential savings were calculated based on normalizing the distribution of weekend discharges to the previous weekday. This results in potential saving for all 5272 infants in charges totaling $1 569 405 using an average of $2500 charged per day for hospital, physician, and home care needs. This calculation yields a total savings of 627 days of hospitalization.
The ALOS was compared by day of the week of discharge and revealed an overall longer ALOS for infants who were discharged on Mondays, compared with those discharged on weekend days (P < .05). The ALOS for all infants discharged on Mondays was 20.5 days, with a Saturday discharge ALOS of 14.9 days and a Sunday discharge ALOS of 13.6 days. The ALOS for term infants discharged on Mondays was 8.4 days, with a Saturday discharge ALOS of 6.9 days and a Sunday discharge ALOS of 6.9 days (P < .05). The ALOS for preterm infants discharged on Mondays was 27.9 days, with a Saturday discharge ALOS of 21.3 days and a Sunday discharge ALOS of 19.2 days, as shown inTable 2.
The timing of ICN discharge requires the resolution of many medical and social issues for the neonate and his or her family. Theoretically, none of these factors should be altered by the day of the week. There are many potential benefits to discharging infants at the earliest safe date. In a randomized study of preterm infants, Gibson et al4 demonstrated that with enhanced outpatient case management, the earlier discharge group had fewer readmissions and improved weight gain at home.3 Similarly, other studies reporting earlier discharge showed that this can be done safely and effectively through effective case management and planning.5–8 The advantages of earlier discharge of infants from an ICN are multifactorial. These include increased rate of weight gain in the home environment, enhanced mother–infant bonding, decreased incidence of iatrogenic illness, and lower cost associated with shorter length of stay readmissions and improved weight gain at home.2
In this study we evaluated patterns of births and discharge of infants from many ICNs in several different geographic regions. If patients were born in a nonrandom manner, and if discharges were optimized, nursery discharges should occur evenly on each day of the week. Furthermore, if infants were sent home in a nonrandom manner, based on parental convenience, there would most likely be more weekend day discharges. There may be some exceptions to this, such as parents who work on weekends or who may not be able to bring their infants home on weekend days because of religious restrictions on travel. This study shows, however, that infants who are admitted to an ICN are neither born nor discharged in a random manner with respect to the day of the week. This pattern persists within both term and preterm subgroups, over a wide range of geographic regions. The issues that produce this nonrandom pattern, therefore, are not unique to a single group or hospital system.
The weekday predominant distribution of both preterm and term births can be explained, in part, by a reluctance to perform elective inductions and cesarean births on the weekends. In addition, the decision to deliver the fetus with a marginally abnormal uterine environment may be delayed until a weekday because of staffing and possible parental issues and desires.9 Because the population studied were all admitted to an ICN, the term population reported was skewed from the general population of deliveries, which may have a more uniform distribution of births and discharges. Because some of the term infants studied were noted to have abnormalities while in utero, however, weekday deliveries may be frequently planned. This may not necessarily represent a prolonged hospitalization for the mother, because she could be home before the scheduled event.
Despite the weekday predominance for births, one would still expect a generally random distribution of discharges throughout the week. Unlike the issues with the day of delivery, a predominance of weekday discharges for both the term and preterm population most likely represents an unnecessary delay in discharge and a prolonged hospitalization for the infant. The comparison of ALOS of infants discharged on Saturday, Sunday, and Monday reveals a significantly longer ALOS for those infants discharged on Monday. This is further evidence that the nonrandom distribution pattern of discharges is representative of unnecessary delays in discharge. The tendency toward weekday discharges may be related to several hospital and physician-related variables, such as decreased weekend staffing levels, familiarity with the infant and family, availability of home care scheduling, outpatient appointments, laboratories, and technical evaluations (eg, computed tomography scans are unavailable), etc. However, parents may be expected to prefer a weekend discharge for work-related reasons. We have speculated that a nonrandom distribution of patients in this study would save over 1.5 million dollars or approximately $300/patient. This calculation is based only on increasing weekend discharges to weekday levels. We did not include other potential savings on the introduction of a more aggressive discharge program, but it is most likely that infants could be discharged earlier during the week in some cases as well. The additional variability in discharge days noted within different regions suggests that other factors exist that delay discharge during weekdays. Some regions for instance demonstrated a tendency to discharge patients predominantly on Fridays or fewer on Mondays. Yet, overall, there is a trend toward more Monday discharges implying that fewer patients are rushed out before the weekend than are detained for a Monday discharge. We speculated that this nonrandom distribution represented a delay in discharge of some patients, rather than earlier discharge of patients, thereby increasing overall length of hospital stay, which is confirmed by a comparison of ALOS of the infants studied.
If fiscal and patient-related issues of a delayed discharge were to be considered, the costs of increasing the availability of necessary services and improving the staffing issues may be very worthwhile. We would conclude, therefore, that an effective multidisciplinary case management approach and discharge planning of infants admitted to an ICN would result in a random distribution of the day of hospital discharge. This may be beneficial to the infant and family and decrease cost.
Dr Kornhauser was supported in part by an educational grant from PHMS, Inc. to the Office of Health Policy and Clinical Outcomes.
- Received April 21, 2000.
- Accepted April 21, 2000.
Reprint requests to (S.M.T.) Jefferson Medical College, 700 College Bldg, 1025 Walnut St, Philadelphia, PA 19107. E-mail:
- ICN =
- intensive care nursery •
- DM =
- disease management •
- ALOS =
- average length of stay
- Committee on Fetus and Newborn
- Kendall M, Stuart A. The Advanced Theory of Statistics, I. New York, NY: Hafner Publishing Co; 1969
- Copyright © 2001 American Academy of Pediatrics