Cognitive Strain
Continuous observation has revealed that nurses’ workdays are characterized by continual reorganization of their work schedules and frequent interruptions.
Belgian (Malchaire 1992) and French (Gadbois et al. 1992; Estryn-Béhar and Fouillot 1990b) studies have revealed that nurses perform 120 to 323 separate tasks during their workday (see table 1). Work interruptions are very frequent throughout the day, ranging from 28 to 78 per workday. Many of the units studied were large, short-term-stay units in which the nurses’ work consisted of a long series of spatially dispersed, short-duration tasks. Planning of work schedules was complicated by the presence of incessant technical innovation, close interdependence of the work of the various staff members and a generally haphazard approach to work organization.
Table 1. Number of separate tasks undertaken by nurses, and interruptions during each shift
Belgium |
France |
France |
|
Authors |
Malchaire 1992* |
Gadbois et al. 1992** |
Estryn-Béhar and |
Departments |
Cardiovascular |
Surgery (S) and |
Ten medical and |
Number of separate |
Morning 120/8 h |
S (day) 276/12 h |
Morning 323/8 h |
Number of |
S (day) 36/12 h |
Morning 78/8 h |
Number of hours of observation: * Morning: 80 h; afternoon: 80 h; night: 110 h. ** Surgery: 238 h; medicine: 220 h. *** Morning : 64 h; afternoon: 80 h; night: 90 h.
Gadbois et al. (1992) observed an average of 40 interruptions per workday, of which 5% were caused by patients, 40% by inadequate transmission of information, 15% by telephone calls and 25% by equipment. Ollagnier and Lamarche (1993) systematically observed nurses in a Swiss hospital and observed 8 to 32 interruptions per day, depending on the ward. On average, these interruptions represented 7.8% of the workday.
Work interruptions such as these, caused by inappropriate information supply and transmission structures, prevent workers from completing all their tasks and lead to worker dissatisfaction. The most serious consequence of this organizational deficiency is the reduction of time spent with patients (see table 2). In the first three studies cited above, nurses spent at most 30% of their time with patients on average. In Czechoslovakia, where multiple-bed rooms were common, nurses needed to change rooms less frequently, and spent 47% of their shift time with patients (Hubacova, Borsky and Strelka 1992). This clearly demonstrates how architecture, staffing levels and mental strain are all interrelated.
Table 2. Distribution of nurses’ time in three studies
Czechoslovakia |
Belgium |
France |
|
Authors |
Hubacova, Borsky and Strelka 1992* |
Malchaire 1992** |
Estryn-Béhar and |
Departments |
5 medical and surgical departments |
Cardiovascular surgery |
10 medical and |
Average time for the main postures and total distance walked by nurses: |
|||
Per cent working |
76% |
Morning 61% |
Morning 74% |
Including stooping, |
11% |
Morning 16% |
|
Standing flexed |
Morning 11% |
||
Distance walked |
Morning 4 km |
Morning 7 km |
|
Per cent working |
Three shifts: 47% |
Morning 38% |
Morning 24% |
Number of observations per shift: * 74 observations on 3 shifts. ** Morning: 10 observations (8 h); afternoon: 10 observations (8 h); night: 10 observations (11 h). *** Morning: 8 observations (8 h); afternoon: 10 observations (8 h); night: 9 observations (10-12 h).
Estryn-Béhar et al. (1994) observed seven occupations and schedules in two specialized medical wards with similar spatial organization and located in the same high-rise building. While work in one ward was highly sectorized, with two teams of a nurse and a nurses’ aide attending half of the patients, there were no sectors in the other ward, and basic care for all patients was dispensed by two nurses’ aides. There were no differences in the frequency of patient-related interruptions in the two wards, but team-related interruptions were clearly more frequent in the ward without sectors (35 to 55 interruptions compared to 23 to 36 interruptions). Nurses’ aides, morning-shift nurses and afternoon-shift nurses in the non-sectorized ward suffered 50, 70 and 30% more interruptions than did their colleagues in the sectorized one.
Sectorization thus appears to reduce the number of interruptions and the fracturing of work shifts. These results were used to plan the reorganization of the ward, in collaboration with the medical and paramedical staff, so as to facilitate sectorization of the office and the preparation area. The new office space is modular and easily divided into three offices (one for physicians and one for each of the two nursing teams), each separated by sliding glass partitions and furnished with at least six seats. Installation of two counters facing each other in the common preparation area means that nurses who are interrupted during preparation can return and find their materials in the same position and state, unaffected by their colleagues’ activities.
Reorganization of work schedules and technical services
Professional activity in technical departments is much more than the mere sum of tasks associated with each test. A study conducted in several nuclear medicine departments (Favrot-Laurens 1992) revealed that nuclear medicine technicians spend very little of their time performing technical tasks. In fact, a significant part of technicians’ time was spent coordinating activity and workload at the various workstations, transmitting information and making unavoidable adjustments. These responsibilities stem from technicians’ obligation to be knowledgeable about each test and to possess essential technical and administrative information in addition to test-specific information such as time and injection site.
Information processing necessary for the delivery of care
Roquelaure, Pottier and Pottier (1992) were asked by a manufacturer of electroencephalography (EEG) equipment to simplify the use of the equipment. They responded by facilitating the reading of visual information on controls which were excessively complicated or simply unclear. As they point out, “third-generation” machines present unique difficulties, due in part to the use of visual display units packed with barely legible information. Deciphering these screens requires complex work strategies.
On the whole, however, little attention has been paid to the need to present information in a manner that facilitates rapid decision-making in health care departments. For example, the legibility of information on medicine labels still leaves much to be desired, according to one study of 240 dry oral and 364 injectable medications (Ott et al. 1991). Ideally, labels for dry oral medication administered by nurses, who are frequently interrupted and attend several patients, should have a matte surface, characters at least 2.5 mm high and comprehensive information on the medication in question. Only 36% of the 240 medications examined satisfied the first two criteria, and only 6% all three. Similarly, print smaller than 2.5 mm was used in 63% of labels on the 364 injectable medications.
In many countries where English is not spoken, machine control panels are still labelled in English. Patient-chart software is being developed in many countries. In France, this type of software development is often motivated by a desire to improve hospital management and undertaken without adequate study of the software’s compatibility with actual working procedures (Estryn-Béhar 1991). As a result, the software may actually increase the complexity of nursing, rather than reduce cognitive strain. Requiring nurses to page through multiple screens of information to obtain the information they need to fill a prescription may increase the number of errors they make and memory lapses they suffer.
While Scandinavian and North American countries have computerized much of their patient records, it must be borne in mind that hospitals in these countries benefit from a high staff-to-patient ratio, and work interruptions and constant reshuffling of priorities are therefore less problematic there. In contrast, patient-chart software designed for use in countries with lower staff-to-patient ratios must be able to easily produce summaries and facilitate reorganization of priorities.
Human error in anaesthesia
Cooper, Newbower and Kitz (1984), in their study of the factors underlying errors during anaesthesia in the United States, found equipment design to be crucial. The 538 errors studied, largely drug administration and equipment problems, were related to the distribution of activities and the systems involved. According to Cooper, better design of equipment and monitoring apparatus would lead to a 22% reduction in errors, while complementary training of anaesthesiologists, using new technologies such as anaesthesia simulators, would lead to a 25% reduction. Other recommended strategies focus on work organization, supervision and communications.
Acoustic alarms in operating theatres and intensive-care units
Several studies have shown that too many types of alarms are used in operating theatres and intensive-care units. In one study, anaesthetists identified only 33% of alarms correctly, and only two monitors had recognition rates exceeding 50% (Finley and Cohen 1991). In another study, anaesthetists and anaesthesia nurses correctly identified alarms in only 34% of cases (Loeb et al. 1990). Retrospective analysis showed that 26% of nurses’ errors were due to similarities in alarm sounds and 20% to similarities in alarm functions. Momtahan and Tansley (1989) reported that recovery-room nurses and anaesthetists correctly identified alarms in only 35% and 22% of cases respectively. In another study by Momtahan, Hétu and Tansley (1993), 18 physicians and technicians were able to identify only 10 to 15 of 26 operating-theatre alarms, while 15 intensive-care nurses were able to identify only 8 to 14 of 23 alarms used in their unit.
De Chambost (1994) studied the acoustic alarms of 22 types of machines used in an intensive-care unit in the Paris region. Only the cardiogram alarms and those of one of the two types of automated-plunger syringes were readily identified. The others were not immediately recognized and required personnel first to investigate the source of the alarm in the patient’s room and then return with the appropriate equipment. Spectral analysis of the sound emitted by eight machines revealed significant similarities and suggests the existence of a masking effect between alarms.
The unacceptably high number of unjustifiable alarms has been the object of particular criticism. O’Carroll (1986) characterized the origin and frequency of alarms in a general intensive-care unit over three weeks. Only eight of 1,455 alarms were related to a potentially fatal situation. There were many false alarms from monitors and perfusion pumps. There was little difference between the frequency of alarms during the day and night.
Similar results have been reported for alarms used in anaesthesiology. Kestin, Miller and Lockhart (1988), in a study of 50 patients and five commonly used anaesthesia monitors, reported that only 3% indicated a real risk for the patient and that 75% of alarms were unfounded (caused by patient movement, interference and mechanical problems). On average, ten alarms were triggered per patient, equivalent to one alarm every 4.5 minutes.
A common response to false alarms is simply to disable them. McIntyre (1985) reported that 57% of Canadian anaesthetists admitted deliberately inactivating an alarm. Obviously, this could lead to serious accidents.
These studies underscore the poor design of hospital alarms and the need for alarm standardization based on cognitive ergonomics. Both Kestin, Miller and Lockhart (1988) and Kerr (1985) have proposed alarm modifications that take into account risk and the expected corrective responses of hospital personnel. As de Keyser and Nyssen (1993) have shown, the prevention of human error in anaesthesia integrates different measures—technological, ergonomic, social, organizational and training.
Technology, human error, patient safety and perceived psychological strain
Rigorous analysis of the error process is very useful. Sundström-Frisk and Hellström (1995) reported that equipment deficiencies and/or human error were responsible for 57 deaths and 284 injuries in Sweden between 1977 and 1986. The authors interviewed 63 intensive-care-unit teams involved in 155 incidents (“near-accidents”) involving advanced medical equipment; most of these incidents had not been reported to authorities. Seventy typical “near-accident” scenarios were developed. Causal factors identified included inadequate technical equipment and documentation, the physical environment, procedures, staffing levels and stress. The introduction of new equipment may lead to accidents if the equipment is poorly adapted to users’ needs and is introduced in the absence of basic changes in training and work organization.
In order to cope with forgetfulness, nurses develop several strategies for remembering, anticipating and avoiding incidents. They do still occur and even when patients are unaware of errors, near-accidents cause personnel to feel guilty. The article "Case Study: Human Error and Critical Taks" deals with some aspects of the problem.
Emotional or Affective Strain
Nursing work, especially if it forces nurses to confront serious illness and death, can be a significant source of affective strain, and may lead to burn-out, which is discussed more fully elsewhere in this Encyclopaedia. Nurses’ ability to cope with this stress depends on the extent of their support network and their possibility to discuss and improve patients’ quality of life. The following section summarizes the principal findings of Leppanen and Olkinuora’s (1987) review of Finnish and Swedish studies on stress.
In Sweden, the main motivations reported by health professionals for entering their profession were the “moral calling” of the work, its usefulness and the opportunity to exercise competence. However, almost half of nurses’ aides rated their knowledge as inadequate for their work, and one-quarter of nurses, one-fifth of registered nurses, one-seventh of physicians and one-tenth of head nurses considered themselves incompetent at managing some types of patients. Incompetence in managing psychological problems was the most commonly cited problem and was particularly prevalent among nurses’ aides, although also cited by nurses and head nurses. Physicians, on the other hand, consider themselves competent in this area. The authors focus on the difficult situation of nurses’ aides, who spend more time with patients than the others but are, paradoxically, unable to inform patients about their illness or treatment.
Several studies reveal the lack of clarity in delineating responsibilities. Pöyhönen and Jokinen (1980) reported that only 20% of Helsinki nurses were always informed of their tasks and the goals of their work. In a study conducted in a paediatric ward and an institute for disabled persons, Leppanen showed that the distribution of tasks did not allow nurses enough time to plan and prepare their work, perform office work and collaborate with team members.
Responsibility in the absence of decision-making power appears to be a stress factor. Thus, 57% of operating-room nurses felt that ambiguities concerning their responsibilities aggravated their cognitive strain; 47% of surgical nurses reported being unfamiliar with some of their tasks and felt that patients’ and nurses’ conflicting expectations were a source of stress. Further, 47% reported increased stress when problems occurred and physicians were not present.
According to three European epidemiological studies, burn-out affects approximately 25% of nurses (Landau 1992; Saint-Arnaud et al. 1992; Estryn-Béhar et al. 1990) (see table 3 ). Estryn-Béhar et al. studied 1,505 female health care workers, using a cognitive strain index that integrates information on work interruptions and reorganization and an affective strain index that integrates information on work ambience, teamwork, congruity of qualification and work, time spent talking to patients and the frequency of hesitant or uncertain responses to patients. Burn-out was observed in 12% of nurses with low, 25% of those with moderate and 39% of those with high cognitive strain. The relationship between burn-out and affective strain increases was even stronger: burn-out was observed in 16% of nurses with low, 25% of those with moderate and 64% of those with high affective strain. After adjustment by logistic multivariate regression analysis for social and demographic factors, women with a high affective strain index had an odds ratio for burn-out of 6.88 compared to those with a low index.
Table 3. Cognitive and affective strain and burn-out among health workers
Germany* |
Canada** |
France*** |
|
Number of subjects |
24 |
868 |
1,505 |
Method |
Maslach Burn-out |
Ilfeld Psychiatric |
Goldberg General |
High emotional |
33% |
20% |
26% |
Degree of burn-out, |
Morning 2.0; |
Morning 25%; |
|
Percentage suffering |
Cognitive and |
Cognitive strain: |
* Landau 1992. ** Saint Arnand et. al. 1992. *** Estryn-Béhar et al. 1990.
Saint-Arnaud et al. reported a correlation between the frequency of burn-out and the score on their composite cognitive and affective strain index. Landau’s results support these findings.
Finally, 25% of 520 nurses working in a cancer treatment centre and a general hospital in France were reported to exhibit high burn-out scores (Rodary and Gauvain-Piquard 1993). High scores were most closely associated with a lack of support. Feelings that their department did not regard them highly, take their knowledge of the patients into account or put the highest value on their patients’ quality of life were reported more frequently by nurses with high scores. Reports of being physically afraid of their patients and unable to organize their work schedule as they wished were also more frequent among these nurses. In light of these results, it is interesting to note that Katz (1983) observed a high suicide rate among nurses.
Impact of workload, autonomy and support networks
A study of 900 Canadian nurses revealed an association between workload and five indices of cognitive strain measured by the Ilfeld questionnaire: the global score, aggression, anxiety, cognitive problems and depression (Boulard 1993). Four groups were identified. Nurses with a high workload, high autonomy and good social support (11.76%) exhibited several stress-related symptoms. Nurses with a low workload, high autonomy and good social support (35.75%) exhibited the lowest stress. Nurses with high workload, little autonomy and little social support (42.09%) had a high prevalence of stress-related symptoms, while nurses with a low workload, little autonomy and little social support (10.40%) had low stress, but the authors suggest that these nurses may experience some frustration.
These results also demonstrate that autonomy and support, rather than moderating the relationship between workload and mental health, act directly on workload.
Role of head nurses
Classically, employee satisfaction with supervision has been considered to depend on the clear definition of responsibilities and on good communication and feedback. Kivimäki and Lindström (1995) administered a questionnaire to nurses in 12 wards of four medical departments and interviewed the wards’ head nurses. Wards were classified into two groups on the basis of the reported level of satisfaction with supervision (six satisfied wards and six dissatisfied wards). Scores for communication, feedback, participation in decision-making and the presence of a work climate that favours innovation were higher in “satisfied” wards. With one exception, head nurses of “satisfied” wards reported conducting at least one confidential conversation lasting one to two hours with each employee annually. In contrast, only one of the head nurses of the “dissatisfied” wards reported this behaviour.
Head nurses of the “satisfied” wards reported encouraging team members to express their opinions and ideas, discouraging team members from censuring or ridiculing nurses who made suggestions, and consistently attempting to give positive feedback to nurses expressing different or new opinions. Finally, all the head nurses in “satisfied” wards, but none of the ones in “dissatisfied” ones, emphasized their own role in creating a climate favourable to constructive criticism.
Psychological roles, relationships and organization
The structure of nurses’ affective relationships varies from team to team. A study of 1,387 nurses working regular night shifts and 1,252 nurses working regular morning or afternoon shifts revealed that shifts were extended more frequently during night shifts (Estryn-Béhar et al. 1989a). Early shift starts and late shift ends were more prevalent among night-shift nurses. Reports of a “good” or “very good” work ambience were more prevalent at night, but a “good relationship with physicians” was less prevalent. Finally, night-shift nurses reported having more time to talk to patients, although that meant that worries and uncertainties about the appropriate response to give patients, also more frequent at night, were harder to bear.
Büssing (1993) revealed that depersonalization was greater for nurses working abnormal hours.
Stress in physicians
Denial and suppression of stress are common defence mechanisms. Physicians may attempt to repress their problems by working harder, distancing themselves from their emotions or adopting the role of a martyr (Rhoads 1977; Gardner and Hall 1981; Vaillant, Sorbowale and McArthur 1972). As these barriers become more fragile and adaptive strategies break down, bouts of anguish and frustration become more and more frequent.
Valko and Clayton (1975) found that one-third of interns suffered severe and frequent episodes of emotional distress or depression, and that one-quarter of them entertained suicidal thoughts. McCue (1982) believed that a better understanding of both stress and reactions to stress would facilitate physician training and personal development and modify societal expectations. The net effect of these changes would be an improvement in care.
Avoidance behaviours may develop, often accompanied by a deterioration of interpersonal and professional relationships. At some point, the physician finally crosses the line into a frank deterioration of mental health, with symptoms which may include substance abuse, mental illness or suicide. In yet other cases, patient care may be compromised, resulting in inappropriate examinations and treatment, sexual abuse or pathological behaviour (Shapiro, Pinsker and Shale 1975).
A study of 530 physician suicides identified by the American Medical Association over a five-year period found that 40% of suicides by female physicians and less than 20% of suicides by male physicians occurred in individuals younger than 40 years (Steppacher and Mausner 1974). A Swedish study of suicide rates from 1976 to 1979 found the highest rates among some of the health professions, compared to the overall active population (Toomingas 1993). The standardized mortality ratio (SMR) for female physicians was 3.41, the highest value observed, while that for nurses was 2.13.
Unfortunately, health professionals with impaired mental health are often ignored and may even be rejected by their colleagues, who attempt to deny these tendencies in themselves (Bissel and Jones 1975). In fact, slight or moderate stress is much more prevalent among health professionals than are frank psychiatric disorders (McCue 1982). A good prognosis in these cases depends on early diagnosis and peer support (Bitker 1976).
Discussion groups
Studies on the effect of discussion groups on burn-out have been undertaken in the United States. Although positive results have been demonstrated (Jacobson and MacGrath 1983), it should be noted that these have been in institutions where there was sufficient time for regular discussions in quiet and appropriate settings (i.e., hospitals with high staff-patient ratios).
A literature review of the success of discussion groups has shown these groups to be valuable tools in wards where a high proportion of patients are left with permanent sequelae and must learn to accept modifications in their lifestyle (Estryn-Béhar 1990).
Kempe, Sauter and Lindner (1992) evaluated the merits of two support techniques for nurses near burn-out in geriatrics wards: a six-month course of 13 professional counselling sessions and a 12-month course of 35 “Balint group” sessions. The clarification and reassurance provided by the Balint group sessions were effective only if there was also significant institutional change. In the absence of such change, conflicts may even intensify and dissatisfaction increase. Despite their impending burn-out, these nurses remained very professional and sought ways of carrying on with their work. These compensatory strategies had to take into account extremely high workloads: 30% of nurses worked more than 20 hours of overtime per month, 42% had to cope with understaffing during more than two-thirds of their working hours and 83% were often left alone with unqualified personnel.
The experience of these geriatrics nurses was compared to that of nurses in oncology wards. Burnout score was high in young oncology nurses, and decreased with seniority. In contrast, burnout score among geriatrics nurses increased with seniority, attaining levels much higher than those observed in oncology nurses. This lack of decrease with seniority is due to the characteristics of the workload in geriatrics wards.
The need to act on multiple determinants
Some authors have extended their study of effective stress management to organizational factors related to affective strain.
For example, analysis of psychological and sociological factors was part of Theorell’s attempt to implement case-specific improvements in emergency, paediatric and juvenile psychiatry wards (Theorell 1993). Affective strain before and after the implementation of changes was measured through the use of questionnaires and the measurement of plasma prolactin levels, shown to mirror feelings of powerlessness in crisis situations.
Emergency-ward personnel experienced high levels of affective strain and frequently enjoyed little decisional latitude. This was attributed to their frequent confrontation with life-and-death situations, the intense concentration demanded by their work, the high number of patients they frequently attended and the impossibility of controlling the type and number of patients. On the other hand, because their contact with patients was usually short and superficial, they were exposed to less suffering.
The situation was more amenable to control in paediatric and juvenile psychiatry wards, where schedules for diagnostic procedures and therapeutic procedures were established in advance. This was reflected by a lower risk of overwork compared to emergency wards. However, personnel in these wards were confronted with children suffering from serious physical and mental disease.
Desirable organizational changes were identified through discussion groups in each ward. In emergency wards, personnel were very interested in organizational changes and recommendations concerning training and routine procedures—such as how to treat rape victims and elderly patients with no relations, how to evaluate work and what to do if a called physician doesn’t arrive—were formulated. This was followed by the implementation of concrete changes, including the creation of the position of head physician and the ensuring of the constant availability of an internist.
The personnel in juvenile psychiatry were primarily interested in personal growth. Reorganization of resources by the head physician and the county allowed one-third of the personnel to undergo psychotherapy.
In paediatrics, meetings were organized for all the personnel every 15 days. After six months, social support networks, decisional latitude and work content all had improved.
The factors identified by these detailed ergonomic, psychological and epidemiological studies are valuable indices of work organization. Studies which focus on them are quite different from in-depth studies of multi-factor interactions and instead revolve around the pragmatic characterization of specific factors.
Tintori and Estryn-Béhar (1994) identified some of these factors in 57 wards of a large hospital in the Paris region in 1993. Shift overlap of more than 10 minutes was present in 46 wards, although there was no official overlap between the night and morning shifts in 41 wards. In half the cases, these information communication sessions included nurses’ aides in all three shifts. In 12 wards, physicians participated in the morning-afternoon sessions. In the three months preceding the study, only 35 wards had held meetings to discuss patients’ prognoses, discharges and patients’ understanding of and reaction to their illnesses. In the year preceding the study, day-shift workers in 18 wards had received no training and only 16 wards had dispensed training to their night-shift workers.
Some new lounges were not used, since they were 50 to 85 metres from some of the patients’ rooms. Instead, the personnel preferred holding their informal discussions around a cup of coffee in a smaller but closer room. Physicians participated in coffee breaks in 45 day-shift wards. Nurses’ complaints of frequent work interruptions and feelings of being overwhelmed by their work are no doubt attributable in part to the dearth of seats (less than four in 42 of the 57 wards) and cramped quarters of the nursing stations, where more than nine people must spend a good part of their day.
The interaction of stress, work organization and support networks is clear in studies of the home-care unit of the hospital in Motala, Sweden (Beck-Friis, Strang and Sjöden 1991; Hasselhorn and Seidler 1993). The risk of burn-out, generally considered high in palliative care units, was not significant in these studies, which in fact revealed more occupational satisfaction than occupational stress. Turnover and work stoppages in these units were low, and personnel had a positive self-image. This was attributed to selection criteria for personnel, good teamwork, positive feedback and continuing education. Personnel and equipment costs for terminal-stage cancer hospital care are typically 167 to 350% higher than for hospital-based home care. There were more than 20 units of this type in Sweden in 1993.