Nursing workload assessment in an intensive care unit: A retrospective observational study using the Nursing Activities Score

4.1 Study type and design

Observational, analytical, and retrospective cohort study with the analysis of the electronic health record database. The study was developed in a six-bed ICU in the Central Region of Portugal. The ICU belongs to the Portuguese National Health System and is classified as a Tertiary (level 3) ICU. During the period under analysis, the ICU had a nurse-to-patient ratio of 1:2, 24 h a day, 365 days a year.

The period under analysis was between 1 June–31 August 2020.

This study was executed and reported following Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for Reporting Observational Studies³³ (Supplementary File 1) and took into account the data documented in patients' electronic health records.

4.2 Sample/participants

Applying the inclusion and exclusion criteria described above, we obtained a total sample of 56 patients in critical condition.

4.3 Ethical considerations

The study was approved by Hospital Council Board and Ethics Committee in July 2020 (Document/Meeting Reference Number: 131 of 23 July 2020) and the data were collected by the principal investigator during September 2020. Confidentiality of the data and anonymity of the participants in the study were always guaranteed. Each patient was coded numerically in chronological order, making it possible to guarantee and maintain the anonymity of the participants throughout the study.

4.4 Data collection

All assessment variables for each patient included in the study were extracted from the electronic health record database.

The sociodemographic variables under analysis were ‘gender’ and ‘age’. The ‘gender’ of each participant was dichotomized into ‘male’ and ‘female’. The ‘age’ was assessed in years considering each participant's date of birth.

The clinical and ICU variables were the ‘type of admission’, ‘reason for hospitalization’, ‘diagnoses’, ‘outcome’, ‘length of stay’, and ‘occupancy rate’. The ‘type of admission’ of each participant was categorized as ‘emergency service’; ‘operating room’; ‘internal transfer’; and ‘external transfer’. The participants presented different (multi)organic problems, thus the ‘reason for hospitalization’ was dichotomized into ‘medical problems’ and ‘surgical problems’. The ‘diagnoses’ were coded according to the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM). The participants' ‘outcome’ was categorized as ‘internal transfer’; ‘external transfer’; or ‘death’. The ‘occupancy rate’ was calculated according to the formula ‘Occupancy rate = [(number of hospitalized participants/ICU capacity) × 100]’. The ‘length of stay’ was counted in days. The ‘point zero’ was the time of admission. The ‘endpoint’ was the time of discharge.

Finally, the ‘nursing workload’ was assessed according to the Portuguese version of the NAS.^{23, 25}

4.5 Validity and reliability/rigour

The original version of NAS²² was validated in a study developed in 99 ICU in 15 countries and the results indicated that it explains 80.8% of the nursing time (vs. 43.3% in TISS-28).^{15, 22, 23, 34}

The NAS includes seven major categories and 23 items. Each item has a specific score based on the real-time assessment of nursing activities length. The sum of the 23 items ranges between 0 and 177%. The total score attributed to each participant results from the sum of the scores of the different items and corresponds to the participants' direct and indirect care needs. The final score represents the amount of time each participant spent (in terms of nursing care) in the last 24 h. Therefore, a total score of 100.0% indicates the work of one nurse over a 24-h period.^{15, 22, 23, 25, 34}

The concurrent validity of the Portuguese version of the NAS was demonstrated by the statistically significant correlation between NAS and TISS 28 (r = .678; p = .000), as well as the result of multiple linear regression (r² = .303, p = .000). The convergent validity was demonstrated by the statistically significant correlation between NAS and SAPS II (r = .542, p = .000), as well as the result of multiple linear regression (r² = .079, p = .000).^{23, 25}

The original version of NAS²² and the Portuguese version of the NAS^{23, 25} are available in Supplementary Files 2 and 3.

4.6 Data analysis

The data were analysed using the IBM® Statistical Package for the Social Sciences Software version 27.0 and Microsoft® Excel 365.

Descriptive statistics were used to analyse the demographic and clinical variables that characterize the study sample, with the presentation of the results in terms of absolute frequencies and respective percentages.

For the occupancy rate, length of stay, and nursing workload, we chose to use measures of central tendency, presenting the average, the SD as well as the median, the interquartile range (IQR), and the extreme values (minimum and maximum), allowing a greater understanding of the variables under analysis.

4.7 Bias

The biases are directly related to the number, type, and complexity of the patients admitted to this ICU during the data collection period. The data were collected by the principal investigator directly from the electronic health record database considering the information recorded by the different professionals during their clinical practice.

The implementation of the (Portuguese version of) NAS was suggested after a 5-year retrospective study conducted in this Portuguese ICU with TISS 28.⁵

The study was developed before an electronic health record update and an ICU major refurbishment that occurred in the final trimester of 2020.

5.1 Sample characterization

In the 92 days (3 months) under analysis, 56 patients (study participants) were hospitalized in this ICU (Table 1). Considering demographic data, 41 participants were male (73.21%) and had an average age of 61.82 (±17.66) years. Considering the type of admission, 25 participants were admitted to the ICU through the emergency service (44.64%), followed by 17 participants admitted from the operating room (30.36%), 11 participants who were transferred from inpatient services (19.64%), and three participants transferred from other institutions (5.36%).

Although the study participants presented (multi)organic failures, the reason for admission to the ICU for 33 participants was mainly because of medical problems (58.93%).

During the study period, 390 diagnoses (ICD-9-CM) were documented, namely: 82 ‘Endocrine, Nutritional and Metabolic Diseases, and Immunity Disorders’ diagnosed in 41 (73.21%) participants; 65 ‘Diseases of the Circulatory System’ diagnosed in 36 (64.29%) participants; 50 ‘Symptoms, Signs and Ill-Defined Conditions’ diagnosed in 36 (64.29%) participants; and 34 ‘Diseases of the Genitourinary System’ diagnosed in 28 (50.00%) participants (in this last ICD-9-CM Chapter, all the 28 participants were diagnosed with ‘Acute Renal Failure’).

Figure 1 shows the number of diagnoses recorded in each Chapter of the ICD-9-CM as well as the number of participants covered by the diagnoses of the different chapters.

Regarding the patients' outcome, 42 participants (75.00%) were discharged from the ICU to other services and/or units of lower care needs/complexity (within the same institution); 13 participants (23.21%) died during the ICU length of stay; and one participant (1.79%) was transferred to another hospital because of complications that required the intervention of an external medical specialty.

5.2 Characterization of the nursing workload

During the period under analysis, there was an average occupancy rate of 73.55% (±16.60%) and 365 assessments of the nursing workload were carried out using the Portuguese version of the NAS.

The median length of stay was 3.77 (IQR 1.83–7.82) days. In this period, the average nursing workload per participant was 67.52 (±10.91) points.

Considering the 56 participants in this study, the average of the first nursing workload assessment per participant was 80.67 (±10.89) points and the average of the last nursing workload assessment per participant was 78.08 (±10.14) points. During the period under analysis, the average nursing workload in the ICU for every 24 h was 267.88 (±76.89) points. Data related to occupancy rate, length of stay, and nursing workload are available in Table 2.

Figure 2 shows the variation of the nursing workload in the ICU for every 24 h during the 92 days under analysis.

Figure 3 shows the correlation between the number of participants and the nursing workload at the ICU for each 24 h.

6.1 Nursing workload

The average nursing workload per participant admitted to the ICU was 67.52 (±10.91) points, which shows that each participant needed an average of 16.21 (±2.62) h of nursing care per day, during the length of ICU stay.

These data are similar to the results reported in Brazilian ICUs,^{18, 24} Spanish ICUs,³⁵ and Italian ICUs.^{28, 36} Lower results were reported in a study developed in four Portuguese ICUs^{23, 25} and in three Brazilian ICUs,^{31, 37, 38} probably because of the ICUs' heterogeneity and the characteristics of the patients admitted in those units. Higher results were reported in a study developed in six Iran ICUs³² were the authors also highlighted significant difference between the workload experienced by nurses on different work shifts (morning > evening > night).

Our data also follow the trend of the results reported by Lucchini et al.²⁸ in Italian ICUs, where the median nursing workload per hospitalized person was 68.1 (IQR 58.3–76.6) points.

6.2 Nursing workload in the first 24 h

Analysing the data related to the first 24 h in ICU, the average of the nursing workload per participant rises to 80.67 (±10.89) points, which shows that the care provided in the first 24 h ‘consumes’ an average of 19.36 (±2.61) h of nursing care.

This increased nursing workload in the first 24 h of ICU stays (compared with the average nursing workload per participant during the length of stay) was already described in a study developed in four Portuguese ICUs.^{23, 25}

A multicentre study developed by Padilha et al.³⁴ with 758 participants in 19 ICUs of 7 countries recorded an average nursing workload per participant in the first 24 h of hospitalization of 72.81 (±31.10) points. Although the global average is lower than the average reported in our study, the values found vary between 44.46 (±13.00) points in the Spanish ICUs and 101.81 (±31.30) points in the Norwegian ICUs.

In the study developed by Lucchini et al.²⁸ in Italian ICUs, there was also a higher workload per participant in the first 24 h of UCI stays compared with the rest of the hospitalization period, with a median of 69.8 (IQR 56.2–82.9) points in the first assessment.

6.3 Nursing workload in the last 24 h

Considering the data related to the last 24 h in ICU, the average of the nursing workload per participant is 78.08 (±10.14) points, which shows that the care performed in the last day of ICU stay ‘consumes’ an average of 18.74 (±2.43) h of nursing care.

Although a Sweden study highlighted that the discharge of patients from the ICU to a general ward represents a challenging transition of care,³⁹ we did not find any reference that specifically discussed the nursing workload generated in the preparation and execution of this process. In our study, the nursing workload in the last 24 h was higher than the average of the remaining days of hospitalization (except for admission day) most likely because of the ‘administrative and management activities’ related to the ‘patient discharge procedures’ whose score was reflected in the assessment instrument used.

6.4 Variation of nursing workload in the ICU

The nursing workload in the ICU for every 24 h varied over the study period with no specific pattern but with a direct correlation with the occupancy rate.

According to National Regulations, the calculation for safe ICU nurse staffing should take into account the typology of each unit and patients' (clinical) characteristics.¹⁹ Considering the number, type, and complexity of the patients admitted in this ICU (and the typology and characteristics of this ICU by itself), our data showed that there were 32 days (35.78%) where the nursing workload assessed by NAS was higher than the available human recourses, overloading nurse staffing, and with the risk of compromising patient safety.

The task force of the World Federation of Societies of Intensive and Critical Care Medicine propose a categorization of ICUs in three levels, based on a variety of parameters that reflect the potential of the unit to provide excellent and expert care to the most acute seriously ill patients.⁴⁰

Based on that classification, ‘Ordem dos Enfermeiros’ (National Nursing Council) recommended that: ‘Primary (level 1) ICUs’ should have a nurse to patient ratio of 1:3; ‘Secondary (level 2) ICUs’ should have a nurse to patient ratio of 1:2; and ‘Tertiary (level 3) ICUs’ should have a nurse to patient ratio of 1:1.¹⁹

Thus, our data showed that the implementation of those recommendations/ratios may contribute to safer nurse staffing (despite of nursing workload variations reported during the study period) and may improve patient safety in this Tertiary (level 3) Portuguese ICU.

However, some authors highlight the need to look preferentially at the nursing workload that each participant generate²⁹ and at the specificities of each shift,^{27, 32} instead of calculating nursing ratios according to occupancy rates, average nursing workload per participant, and/or the total 24-h nursing workload in ICU.