Original Article

Characterization of Neonatal Personnel Time Inputs and Prediction From Clinical Variables ¾ A Time and Motion Study

¹Department of Neonatology, Beth Israel Deaconess Medical Center, Boston, MA, USA

²Harvard Newborn Medicine Program, Harvard Medical School, Boston, MA, USA

³Department of Maternal and Child Health, Harvard School of Public Health, Boston, MA, USA

Correspondence to: John A. F. Zupancic, MD, ScD, FRCPC, Department of Neonatology, Beth Israel Deaconess Medical Center, Room 318, Rose Building, 330 Brookline Avenue, Boston, MA 02458, USA

Abstract

OBJECTIVE: To characterize and predict personnel time inputs to neonatal intensive care using infant characteristics from chart review.

STUDY DESIGN: For 12 hours each day, observers timed all direct care, charting, discussions, and procedures for 154 infants. Time inputs were correlated with 40 infant characteristics and resource markers, as well as the Score for Neonatal Acute Physiology (SNAP) for that day of care.

RESULTS: Nurses accounted for 76%, respiratory therapists 8%, fellows 5%, nurse practitioners 7% and attendings 5% of total time invested in patient care. Nurses and respiratory therapists spent proportionately more time in direct patient care. In regression models, a limited number of variables explained 36% of the variance in time input per patient for respiratory therapists (p<0.0001), 42% for nurses (p<0.0001), and 23% for physicians and nurse practitioners (p<0.0001).

CONCLUSIONS: Total labor inputs can be accurately predicted through the use of a limited number of clinical characteristics. This technique should be routinely employed to improve the accuracy of economic evaluations. Nursing accounts for the majority of time invested in neonatal care. Improved efficiency in neonatology is thus most likely to be generated by interventions that reduce direct nursing time. Journal of Perinatology (2002) 22, 658-663 doi:10.1038/sj.jp.7210821

Economic evaluation of neonatal intensive care technologies has been hampered by limitations in the tools for measurement of hospital costs. In most US studies, costs are calculated from charges in administrative billing databases. These databases typically provide careful tallies of ancillary utilization (i.e., costs for laboratory, radiology, and pharmacy), but lump all other daily charges into a "per diem" rate. The per diem typically varies only according to whether an infant is in intensive ("level 3") or intermediate ("level 2") care. This process thus ignores all of the variability in personnel costs between patients in the same general level of care. Such an omission is not trivial: nursing costs alone account for between 50% and 70% of total neonatal intensive care costs.^1,2,3,4 Reliable economic evaluation therefore demands that personnel costs be considered at a more detailed level than is current practice.

As it is not feasible to measure personnel labor inputs each time an economic study is undertaken, we sought to provide researchers with a tool to predict daily time inputs from information in the hospital chart. These measurements of clinical activity are also the first published exploration of the relative contribution of various task types for neonatal nurses, respiratory therapists, nurse practitioners, and physicians.

METHODS

Subjects and Setting

Participants were patients and staff in the neonatal intensive care unit (NICU) at the Beth Israel Deaconess Medical Center in Boston, MA. During the study period, this tertiary level unit had approximately 30 beds with 750 admissions annually and provided support for a general and high-risk obstetrical service with 5000 deliveries annually. The staff comprised a total of 57 full-time equivalent nurses, of whom 11 to 12 were scheduled for each 12-hour shift. Nursing sign-out occurred twice per day, at 7 AM and 7 PM.

Each patient was assigned either to a fellow or to a nurse practitioner for primary care, with an attending physician supervising. This NICU has no pediatric residents. Formal rounds occurred once per day, in the morning, and were attended by the fellow, nurse practitioner, bedside nurse, attending physician, and respiratory therapist, in addition to other personnel not included in the study, such as nutritionists and social workers. Rounds to transfer care to the covering night staff occurred once per day in late afternoon.

Data Collection

During July and August of 1998 and 1999, student observers in the NICU identified one or two infants per day for study participation. All infants were eligible, with the exception of those who were excluded by their attending physicians because of moribund condition or parental distress. Observers were instructed to remain seated near enough to the patient to observe all interactions with personnel, but to avoid providing any obstacle to care. Observers used a stopwatch to time all interactions between the patient and personnel, including registered nurses, nurse practitioners, respiratory therapists, subspecialty fellows, and attending neonatologists. Interactions were subdivided into (1) direct patient care (including any activity in which there was direct contact between the clinician and patient, except when a procedure was performed), (2) rounds and discussion, (3) charting and note writing, (4) procedures, and (5) preparation time for medications and feedings. When nurses left the bedside for charting or preparation, observers requested the reason and timed the absence. Infants were observed for a total of 8 to 12 hours each, from morning to evening rounds. Times for each interaction were recorded on a standardized reporting form along with the category and personnel involved. If more than one type of personnel were involved, as they were during some procedures and during rounds, then time was counted separately for each person present.

Retrospective chart review of each observed infant's chart was conducted for the 24-hour period corresponding to the period of observation, beginning at the preceding midnight. Abstractors calculated the original Score for Neonatal Acute Physiology (SNAP)⁵ for the day of observation and scanned the chart for 40 infant characteristics and resource markers derived from the Neonatal Therapeutic Intensity Scoring System (NTISS).⁶

The study was approved by the institutional review board of the Beth Israel Deaconess Medical Center. Nurses could decline participation when approached by the observer, although none availed themselves of this option.

Analysis

Results of patient observation and chart review were entered into a spreadsheet by the student observers (Excel 97, Microsoft Corporation, Redmond, WA). Statistical analysis was completed using the Statistical Analysis System version 6.12 (SAS Institute, Cary, NC) and SPSS 7.5 (SPSS, Chicago, IL). Times inputs were first normalized to a 24-hour period by dividing each time by the proportion of the day over which the infant was observed. Time inputs for each personnel and work category were summarized descriptively. Time inputs for RNs, clinicians (consisting of the total inputs for nurse practitioners, fellows, and attendings), and respiratory therapists were used as dependent variables for ordinary least squares regression. Initially, univariate regression was performed using the following groups of independent variables: infant characteristics, illness severity as measured by SNAP, and resource utilization as recorded from the hospital chart. Items found to be predictive at the 10% level of statistical significance were then entered into a stepwise regression procedure to derive one predictive model for each of the three personnel categories. Dependent variables with highly skewed distributions were log transformed before regression.

RESULTS

Characteristics of Observed Infants

During the study, 154 infants were observed and had chart abstraction completed, for a total of 1235 hours of observation time. Characteristics of these infants are given in Table 1. As shown, there was a reasonable range of resource intensity, with approximately one third requiring mechanical ventilation on the day of observation, whereas two thirds were receiving at least partial nasogastric feeds. The range of birth weights is typical of a tertiary NICU, with a mean of 1638 g and a range of 460 to 4370 g. The mean SNAP score of 4 corresponds to mild severity of illness. Infants with moderate to severe illness acuity (SNAP >10) comprised 7% of the sample.

Characterization of Time by Clinical Role and Activity

Registered nurses accounted for 76% of the average time input recorded per infant, whereas attending physicians accrued 5%, nurse practitioners 7%, fellows 5%, and respiratory therapists 8%.

Median time inputs per patient projected per 24 hours (1440 minutes) were 385 minutes for registered nurses, 86 minutes for MDs, fellows and nurse practitioners combined, and 22 minutes for respiratory therapists. The distributions for these time inputs were right skewed, as shown for nursing and clinicians in Figures 1 and 2, respectively. The distribution for respiratory therapists was similar.

Of the total time invested in each patient by all personnel, 59% on average was classified as direct patient care, 16% as rounds and discussion, 15% as charting and note writing, 7% as procedures, and 3% as preparatory activities. Figure 3 shows the breakdown of task types by clinical role. As shown, nurses and respiratory therapists were more involved in direct patient care, whereas physicians (attending and fellows) and nurse practitioners spent proportionately more time per patient in rounds and discussions.

Regression Analysis

Table 2 shows the results of regression analysis for log-transformed personnel time. For nursing time, the model predicted 42% of the variability in time inputs (R²=0.42, p<0.0001) and used eight predictive variables. For physicians and nurse practitioners combined, the model predicted 23% of the variability (R²=0.23, p=0.0001) and used three predictive variables. For respiratory therapists, five variables were predictive of 36% of the variability (R²=0.36, p<0.0001).

DISCUSSION

With ongoing financial constraints on the health care system, clinicians and managers are paying increased attention to economic evaluation as a means to direct resources into the highest yield therapies. The value for money of neonatal intensive care has been a particularly active debate for more than two decades.^7,8,9 This patient population certainly claims a disproportionate share of pediatric in-hospital resources, yet despite the debate, few of its therapies have been subjected to careful cost-effectiveness analysis.^10,11

One important reason for the paucity of prospective economic studies may be a lack of confidence in the methods of costing. As noted earlier, traditional methods of classification of neonatal care are not based on direct patient and staff observation and do not adequately account for differences in workload of personnel caring for critically ill infants. Moreover, they vary substantially in their applicability between clinical centers.¹² Both may result in misleading conclusions for economic evaluation. Indeed, no economic evaluations in the neonatal literature have taken explicit account of personnel time inputs.¹¹

Outside the context of economic evaluation, there have been three previous studies focused on personnel time. In the earliest study, activities of nurses and physicians were recorded using time-lapse photography, but the investigation was performed well before the current era of neonatal intensive care and is unlikely to be applicable given the dramatic changes in management since then.¹³ In a second study, British nurses recorded their own time use while caring for 45 infants.¹⁴ This study was also performed in an era before the widespread use of exogenous surfactant and newer modes of ventilation, was limited to nursing workload, and made no attempt to predict variability of time input beyond three broad classification levels corresponding to infants receiving ventilation, high- and low-dependency care. Finally, the Northern Neonatal Network reported on direct observation of nursing time input in several levels of care.¹⁵ No statistical testing was performed regarding the degree to which these classifications explained variability in time input, and the distribution of time among different tasks was not presented.

In contrast, our study provides over 1200 hours of direct observation of nursing, physician, nurse practitioner, and respiratory therapist workload in 154 neonates cared for in a modern, North American tertiary level NICU. It confirms that the majority of time input is by nursing staff for direct patient care. There is, however, a very broad distribution of time input across patients whose per diem "costs" would otherwise be classified at one of only two broad levels of care, intensive and intermediate.

The regression equations allow prediction of relative workload for an individual patient from variables easily obtained from the hospital chart. For nurses, these variables included markers of illness severity (SNAP score, oxygen) and maturity (birth weight, nasogastric feeds), as well as variables that would directly increase nursing time, such as transfusion, use of parenteral nutrition, or phlebotomy. Physician and nurse practitioner time was predicted by a smaller subset of variables that more closely align with traditional neonatal classifications, including type of mechanical ventilation and nasogastric feeds. As expected, respiratory therapist time was correlated well with type of respiratory support.

The findings will be of interest to investigators interested in improving the accuracy of economic evaluations in neonatal intensive care. They also provide a systematic picture of how time is apportioned between personnel in the NICU, and between tasks. This may facilitate the identification of components of neonatal care with a higher yield for economic evaluation. For example, technologies that decrease direct nursing time are likely to be more cost-effective than those that reduce procedural time for physicians.

Several limitations of the study merit discussion. First, NICUs may vary from each other in both the characteristics of the infants admitted and in how personnel care for those infants. Our study population reflects the distribution of illness severity in a NICU of this size, which includes a large number of growing infants. The illness acuity is therefore relatively low. It is unclear whether the predictive validity of the equations would hold if applied to a much sicker population. Similarly, our use of neonatologists in a supervisory role supplemented by both fellows and nurse practitioners may yield different results than in a unit that relies on house staff. However, the combined time from all staff performing duties delegated from the attending neonatologist should be relatively constant. For this reason, we predicted combined fellow, nurse practitioner, and attending neonatologist workload. Nursing duties are relatively constant across units, and should therefore be less susceptible to such problems.

There is also variability in the elasticity with which the NICU responds to changing demands on its personnel resources. Some units, especially those operating near the limit of personnel availability, may respond to increased work demands by increasing the work output per caregiver. In contrast, other units may maintain the same output per individual, while making more personnel available. These differences would affect the predicted time inputs from the regression equations. Our results were collected over a long enough period of time that the spectrum of unit census and acuity should be represented, but the possibility that other units respond differently at the extremes of personnel work demands cannot be excluded.

A second concern relates to the completeness of observation. The observers were stationed at the bedside. Although they made every attempt to capture all clinical activity related to the patient, and received full cooperation from participating personnel, it is possible that some tasks were not included in the timed totals. This is likely to be less of a problem for the estimates of nursing and respiratory therapist time than for physicians and nurse practitioners, who typically spend more time away from the bedside discussing patients or making notes.

A related concern is that observations were made for only 8 to 12 hours at a time, and then normalized to a 24-hour day. It is possible that the activities during the observation period may have been more or less intense than those for the remainder of the day, so that normalization would lead to over- or underestimation of total time input. We attempted to minimize this problem by observing during different shifts, and by including one of the two nursing sign-outs that occurred between shifts.

Finally, it should be emphasized that the observations were made using the infant as the unit of observation. The findings thus exclude some overhead elements of personnel time use, such as in-service training, committee work, and quality improvement activities. Although these elements would be constant between infants and thus less relevant for economic evaluation, the omission would introduce a source of error if the findings were used for practical management of neonatal staffing or for developing reimbursement mechanisms.

Acknowledgements

The authors gratefully acknowledge the assistance of Shawn Stewart, who assisted with data analysis, and of Ashish Shah, James Godin, and Christine Bumatay, who performed the personnel observations and chart abstraction while participating in the summer student program of the Division of Newborn Medicine, Harvard Medical School. We are also indebted to the nurses, respiratory therapists, staff, and administration of the neonatal intensive care unit of the Beth Israel Deaconess Medical Center for their enthusiastic facilitation of the data collection.

Financial Support

References

1 Khoshnood B, Lee KS, Corpuz M, Koetting M, Hsieh HI, Kim BI. Models for determining cost of care and length of stay in neonatal intensive care units. Int J Technol Assess Health Care 1996; 12: (1) 62-71. MEDLINE

2 Ewald U. What is the actual cost of neonatal intensive care? Int J Technol Assess Health Care 1991; 7: (Suppl 1) 155-61. MEDLINE

3 Rogowski J. Cost-effectiveness of care for very low birth weight infants. Pediatrics 1998; 102: (1 Pt 1) 35-43.

4 Rogowski J. Measuring the cost of neonatal and perinatal care. Pediatrics 1999; 103: (1 Suppl E) 329-5. MEDLINE

5 Richardson DK, Gray JE, McCormick MC, Workman K, Goldmann DA. Score for Neonatal Acute Physiology: a physiologic severity index for neonatal intensive care. Pediatrics 1993; 91: (3) 617-23. MEDLINE

6 Gray JE, Richardson DK, McCormick MC, Workman-Daniels K, Goldmann DA. Neonatal therapeutic intervention scoring system: a therapy-based severity-of-illness index. Pediatrics 1992; 90: (4) 561-7. MEDLINE

7 Zupancic JA, Richardson DK, Lee K, McCormick MC. Economics of prematurity in the era of managed care. Clin Perinatol 2000; 27: (2) 483-97. MEDLINE

8 Richardson DK, Zupancic JA, Escobar GJ, Ogino M, Pursley DM, Mugford M. A critical review of cost reduction in neonatal intensive care: I. The structure of costs. J Perinatol 2001; 21: (2) 107-15.

9 Richardson DK, Zupancic JA, Escobar GJ, Ogino M, Pursley DM, Mugford M. A critical review of cost reduction in neonatal intensive care: II. Strategies for reduction. J Perinatol 2001; 21: (2) 121-7. Article MEDLINE

10 Mugford M. The cost of neonatal care: reviewing the evidence. Soz Praventivmed 1995; 40: (6) 361-8.

11 Zupancic JAF, Richardson DK. Systematic review of neonatal randomized controlled trials reveals paucity of ancillary economic evaluations [Abstract]. Pediatr Res 2001; 49: (4) 364A.

12 The ECSURF (Economic Evaluation of Surfactant) Collaborative Study Group. Limited comparability of classifications of levels of neonatal care in UK units. Arch Dis Child Fetal Neonatal Ed 1998; 78: (3) F179-84.

13 Tyson JE, Clarkson JE, Sinclair JC, Leitch R. Analysis of newborn intensive care by time-lapse photography. Crit Care Med 1981; 9: (11) 780-4.

14 Williams S, Whelan A, Weindling AM, Cooke RW. Nursing staff requirements for neonatal intensive care. Arch Dis Child 1993; 68: (5 Spec No) 534-8. MEDLINE

15 Northern Neonatal Network. Measuring neonatal nursing workload. Arch Dis Child 1993; 68: (5 Spec No) 539-43.

Figures

Figure 1 Distribution of total time input by nurses.

Figure 2 Distribution of total time input by attending and fellow physicians and nurse practitioners, combined.

Figure 3 Time spent on each task type, by clinical role.

Tables

Table 1 Characteristics of Observed Infants

Table 2 Regression Analyses of Personnel Time on Clinical Variables

December 2002, Volume 22, Number 8, Pages 658-663

Table of contents    Previous  Article  Next    [PDF]