- Email: [email protected]
General practitioners' subjective experience of surgery workload
SW. Sci. Med. Vol. 33, No. 2. pp. 161-166, 1991 Printed in Great Britain. All rights reserved
GENERAL
Copyright
PRACTITIONERS’ OF SURGERY
SUBJECTIVE WORKLOAD
0
0277-9536/91 S3.00 + 0.00 1991 Pergamon Press pit
EXPERIENCE
PAUL NORMAN,MIKE F~I-IERand TOBY WALL MRC/ESRC Social and Applied Psychology Unit, Department of Psychology, University of Sheffield, Sheffield SIO 2TN, England Abstract-General practitioners (GPs) in the U.K. frequently cite heavy workload as a major source of work-related stress. However, there is little direct evidence either to support this claim or to suggest whether the source of this stress lies in the amount, or rather the type, of work undertaken. The present paper uses a within-subjects design to address these issues in the surgery context. Over a two-week period three GPs gave ratings of subjective workload following each of their surgeries (N = 20-23). Correlational analysis showed a wide range of surgery characteristics to predict the GPs’ nactions, including the length of surgeries, the number of consultations, the number of re-presentations, and the number of follow-ups and investigations requested. The GPs’ perceptions of their patients (e.g. in terms of their health status and the appropriateness of the consultation), in contrast, did not emerge as significant predictors. The results of multiple regression analyses, however, revealed that most of the first-order effects could be accounted for by surgery length. Thus while the study supported the hypothesised link between the overall level of surgery work demands and the subjective experience of workload, it failed to confirm any consistent additional effects of the type of work involved. It was also found that as the length of surgeries increased, the GPs were more likely to arrange follow-up appointments. The implications of the present findings are therefore two-fold. First, that future research on surgery work demands and the subjective experience of workload must take account of surgery length as a primary predictor; and second, that attention be paid to how GPs cope with heavy workloads. Key words-general
practitioners,
surgery workload, subjective experience, coping strategies
INTRODUCTION
Many general practitioners (GPs) in the U.K. consider themselves to be under stress at work, and point to the amount and nature of their workload as major contributing factors [l]. Three strands of evidence give credence to this claim. First, surveys on GPs’ attitudes to their work have shown that many report a feeling of ‘lack of time’ and that they find their work to be frustrating and stressful [2,3]. For example, in a sample of Aberdeen GPs, 87% reported having experienced periods of exhaustion’ (31. Second, the demands of the GP’s work have been linked to low levels of job satisfaction and psychological well-being [4]. Third, there is objective data to suggest that GPs may be at particular risk of illnesses and other behavioural indicators of stress [5]. For instance, in Scotland, the rate of first admissions for alcohol dependence has been found to be two to seven times higher amongst GPs than comparable controls [6]. Such evidence has prompted other investigators to map the objective work demands faced by GPs [7-g]. A study conducted by the DHSS [lo] of over 1000 GPs found the average GP to be someone who spends almost half his/her time each week conducting nearly 150 surgery consultations. Whilst such descriptive data has fuelled a continuing debate surrounding the acceptability of GPs’ work demands [ll-141, it fails to demonstrate a direct link between heavy work demands and potential negative reactions of the kind which are associated with stress at work. To know that the average GP conducts nearly 150 surgery consultations per week is not informative in itself. It is unclear whether this should be considered to be
high or low, stressful or not. To address the question more directly a minimum requirement would be to show that as the number of consultations vary so too do GPs’ psychological reactions. Implicit in much of the literature on GP’s work demands is the view that it is not only the amount of work that GPs undertake that is stressful, but also the nature of that work. Job stress research generally has shown ‘responsibility for others’ to be a key workrelated stressor [IS]. So, it may the case that when GPs are faced with patients who are under great physical or emotional distress, their awareness of this responsibility is heightened and this may be a source of stress. On the other hand though, there is evidence to suggest that GPs find many of their consultations to be either trivial, unnecessary, or inappropriate and that this leads to a feeling of frustration [2]. This may in turn affect the quality of job performance [16]. Either way, it is likely that GPs’ reactions to their consulting patients will be related to their perceptions of, and psychological reactions to, their surgery workload. For such factors to be taken seriously as potential work-related stressors for GPs, it needs to be shown that GPs’ reactions to their work change as a function of these factors. When faced with heavy surgery work demands, GPs may employ certain coping strategies. For instance it has been suggested that the cutting short of consultations by asking patients back at a later date may be a strategy which is frequently employed by GPs when faced with busy surgeries [17]. In fact, it has been documented that about 60% of consultations are GP-initiated re-presentations [18, 191. However, it may be the case that this response simply
161
PAUL NORMANer al.
162
stores up the work for the future, and that it will be the cause of additional stress when the patients re-present [17]. Thus, this type of consultation may be a particular work-related stressor for GPs. Given that GPs frequently report heavy workload as a stressor, that there is evidence to link this to morbidity and mortality statistics, and that they spend almost half their work-time in surgery consultations, it was decided to conduct a study to explore the determinants of their subjective experience of surgery workload. This could be achieved in one of two main ways. The first is to study the experiences of GPs working under different levels or types of work demand, using an independent subjects research design. This has the advantage of allowing larger scale investigations but, because of the difficulty of holding third factors constant (e.g. practice characteristics) it has interpretational difficulties. The second approach is to adopt a within subjects, or time series, research design in which variations in work demands over time are related to psychological reactions for the same GPs working under otherwise equivalent circumstances. Both approaches are required and complement each other. However, it is the time series approach which is adopted here, where we investigate the relationship of both the amount and type of surgery work undertaken by three GPs to their subjective experience of surgery workload. More specifically, on the basis of previous research, a number of hypotheses were formulated: firstly, that high subjective workload ratings would be related to the number of consultations in a surgery and its length (i.e. the amount of work); secondly, that variations in certain types of consultations (e.g. re-presentations) would be associated with the GPs’ subjective workload ratings over and above that accounted for by the total number of consultations in a surgery (i.e. the type of work); and thirdly, that the GPs’ perceptions of their consulting patients (e.g. of their emotional and physical state, and of the appropriateness of the consultation) would also influence ratings of subjective workload. Finally, it was hypothesised that the GPs may employ certain coping strategies (e.g. referring and arranging follow-up appointments) when faced with busy surgeries. METHOD
The study focus
The study was conducted at a single practice consisting of three male GPs with a combined list size of approximately 7000 patients. The practice is situated in a town in the East Midlands, England. The three GPs completed short questionnaires at the end of each consultation and each surgery over a twoweek period. During this period 733 consultations were conducted in 68 surgeries. Questionnaires were returned for 688 of the consultations (93.9% response rate) and for 67 of the surgeries (98.5% response rate). Surgery characteristics
The three GPs provided detailed data on the characteristics of each surgery they conducted over the two-week study period. Firstly, they recorded the
type of each consultation. These fell into two main categories: presentations and re-presentations. Secondly, they recorded any actions resulting from the consultation. These fell into four main categories: no follow-up arranged, follow-up with the GP arranged, referral, and the requesting of investigations. For further analysis these data were summed for each surgery so that the influence of the make-up of surgeries (in terms of the number of different types of consultations and resulting actions) on the subjective experience of surgery workload could be assessed. In addition, the GPs assessed each patient’s health status. Thus a health status scale was produced for each patient by combining the GPs’ ratings along three dimensions: disability (7 point scale: no disability, slight social, affects work, cannot work, needs help, bed-bound, unconscious), emotional distress and pain/physical distress (4 point scales: none, mild, moderate, severe). These measures were based on scales developed previously [20]. Finally, the GPs rated on 5 point scales (very high, high, medium, low, very low) their perceptions of each patients ‘need’ for the consultation. In short, they rated the urgency (“how urgent was the patient’s need for the consultation?‘), the importance (“how important was it that the patient attend the consultation in terms of an ongoing state of health?‘), the appropriateness (“how appropriate was the consultation in terms of your specific role in the Primary Health Care Team?‘), and the helpfulness (“how helpful do you feel the consultation was for the patient?“) of each consultation. Again, these ratings were combined to produce a consultation appropriateness score. For analysis the above ratings were averaged for each surgery in order to examine whether the GPs’ perceptions of their consultations over the course of a surgery affected their subjective experience of surgery workload. The GPs also indicated the start and end time of each surgery. From this it was possible to calculate the length of each surgery. Surgery workload questionnaire
The questionnaire employed in the present study to assess each GP’s subjective experience of their workload was a slightly modified version of the NASA Task Load Index, which has been developed, tested and used in a range of experimental situations to measure work-related stress [21]. The three GPs therefore rated each surgery in terms of several perceived aspects of their workload [22]. These were termed mental demand [“How much mental and perceptual activity (e.g. searching for information, making decisions, performing calculations) was required? Were your duties easy or demanding?“]; physical demand (“How much physical activity was required? Was the physical demand slack or strenuous?‘); temporal demand (“How much time pressure did you feel under as the result of the rate of consultations? Was the pace slow and leisurely or rapid and frantic?” ); performance (“How successful do you think you were in accomplishing the goals of your duties?‘); efort (“How hard did you have to work (mentally and physically) to accomplish your level of performance?‘); andfncstration (“How much frustration did you feel during the surgery? Were YOU content and relaxed or discouraged and irritated?“).
GPs’
subjective experience of their workload
These reactions were assessed via ten-point rating scales (low/high anchors) covering each of the six aspects, with a 0 therefore indicating low perceived demand and a 9 high perceived demand. Initial analyses revealed scores on these six items to be highly intercorrelated. Combining them to produce an overall workload scale for each GP yielded reasonably high alpha coefficients (see Table I), thus indicating satisfactory levels of internal consistency. Furthermore, the scale was found to relate to the predictor variables in the same way as the individual component items. For the following analyses overall workload scale scores have therefore been used. Data analysis
The data from each GP were analysed as a separate time series, following procedures described by Box and Jenkins [23], and others [24,25]. Since repeated measurement using the same instrument could influence the scores obtained (either through recall of previous responses or through ‘practice’ effects), the first stage of analysis was to examine each time series for the presence of serial dependences and trends. No such effects were found, with the exception of a small and isolated auto-correlation (lag = 1) effect for one GP only. Consequently, there was no need to take account of such effects in subsequent analyses. In addition, given the limited size of the data set, each variable was examined to determine whether it displayed sufficient variation to warrant further investigation. This was found to be the case for all the variables considered for each of the GPs. Analysis then proceeded in two further stages. First, the relationship between each of the independent variables and the dependent variable of subjective workload was tested by calculating product-moment correlations from each GP’s data, thus testing for first-order effects. However, since several of the independent variables were logically and empirically interrelated (e.g. longer surgeries tend to have more consultations, more referrals, etc.), the second stage involved multiple regression analyses. Here, for each GP, length of surgery was entered first into the regression equation on the basis that conceptually and empirically it was the primary predictor of the subjective experience of surgery workload, followed by any other variables explaining additional variance. This therefore tested for the effect of these other surgery characteristics over and above that accounted for by surgery length. However, given the close interdependence of many of the variables considered here, as highlighted above, there is a potential problem of multicollinearity. Since this has implications for the validity of regression techniques by making beta weights unstable [16], it was important to determine its relevance in the present case. To address this two steps were taken. First, correlation matrices of the independent variables for each GP were computed and analysed to determine whether any variables were interrelated to an extent that might invalidate regression analyses. This was not found to be the case. Second, in conducting the regression analyses, SPSSx was employed which calculates the tolerance levels of the independent variables and therefore provides a test SSM 3312-E
163
for whether multicollinearity is a problem. Again, this was not found to be the case. RESULTS
First -order effects
The results of the first-order analyses are presented in Table 1. Five variables were found to stand out as strong predictors of the subjective experience of surgery workload for all three GPs. These were the length of the surgery, the overall number of consultations, the number of re-presentations, the number of consultations when no follow-up was arranged, and the number of requests made for investigations during the course of the surgery. In addition, the number of follow-ups arranged during the course of the surgery predicted the ratings for two of the three GPs. It is evident therefore that the main predictors of the GPs’ subjective workload ratings centred around the objective surgery work demands, both in terms of the amount of work conducted (e.g. surgery length) and its type (e.g. re-presentations). By contrast, the GPs’ perceptions of their consulting patients, in terms of their health status and their ‘need’ for the consultation, were largely unrelated to their subjective workload ratings. In parentheses: it is worth noting that given the centrality of surgery length (or time) as a predictor of the subjective experience of surgery workload, we considered the predictive value of alternative variants of this core construct. Following the suggestions of an anonymous referee, two more variables were created. Both centred on the amount of work involved in relation to time. The first was a ratio score of time in relation to numbers seen and the second was a measure of the extent to which the surgeries over-ran. The over-run measure correlated highly with workload ratings for all three GPs (rs = 0.84, 0.79, 0.67, Ps < 0.001) as did the ratio score for GPl and GP3 (rs = 0.64, 0.51, Ps c 0.01.0.05). However, given that neither of these measures explained variance in addition to surgery length, and being derived directly from this core variable introduce problems Table 1. Zero order correlates of subjective workload ratings for each GP GP: Amount of work Surgery length Number of consultations Type of work Presentations Re-presentations Actions from consultations: No follow-up Follow-up Referrals Investigations Perceptions: Patient health status Consultation appropriateness Mean: SD: Alpha:
Subjective workload ratings I 2 3 0.81*** 0.68’9.
0.77*** 0.76’**
0.72*** 0.54**
0.67*** 0.40.
0.21 0.66***
0.29 0.67***
0.46’ 0.60.. -0.07 0.40.
0.44. 0.53** 0.26 -o.so**
-0.04 0.26
0.32 -0.04
0.35 0.37.
31.70 9.00 0.93
36.39 6.63 0.73
40.19 7.18 0.76
Notes: GPI n = 20, GP2 n = 23, GP3 n = 21. lf c 0.05, l*p < 0.01, l**p < 0.001.
0&O** -0.07 0.08 0.43.
164
PAULNOW
of multicollinearity, they were not considered subsequent analyses. Second-order
in
Table 2. Regression analyses: significant independent predictors of each GP’s subjective experience of their surgery workload (beta values1 GP:
Subjective workload ratings I 2 3
s1ep 1: &Ill
vahes Surgery length
F-value df R Square Stev 2: Beta values Surgery length Investigations F-value
4 R Square
o.s1***
3J.20*** (1.18) 0.66
0.779.. 30.17999 (1.21) 0.59
0.72+** 20.26**’ (ii;)
.
0.709’. -0.36” 25.12*** (2.20) 0.72
Step 3: Beta calues
Surgery length Investigations Number of consultations F-value R Sauare
0.41. -0.34. 0.40. 23.45*** (3.19) 0.79
Notes: GPI n = 20, GP2 n = 23, GP3 n = 21. lP < 0.05, l*P c 0.01. ***F- < 0.001.
Tabk 3. Correlations between the number of consultations in a surgery and the employment of different actions
effects
The second step in identifying the best predictors of each GP’s subjective experience of their surgery workload was to enter those variables which were found to significantly correlate for all three GPs into a series of stepwise regression analyses. Thus for each GP the five predictor variables outlined above were used to predict their ratings of subjective workload, but with surgery length being the tirst variable to be entered into the regression equation. This is clearly essential because, if the number of consultations, re-presentations, follow-ups not arranged, or investigations requested over the course of a surgery are to be taken seriously as potential stressors, it must be demonstrated that their effects exist over and above the fact that they simply occur more frequently in longer surgeries. The results of the regression analyses are presented in Table 2, and point to two main findings. Firstly, surgery length was found to be a significant predictor for each of the three GPs when entered by itself, accounting for between 52% and 66% of the variance in the subjective workload ratings. Secondly, when the other predictor variables were entered into the regression equations only two significant predictors emerged; both for GP2, for whom the number of investigations requested over the course of a surgery accounted for a significant proportion of the variance in subjective workload ratings over and above the effect of surgery length (R square change = 0.13, F change = 8.83, P < O.Ol), as did the overall number of consultations (R square change = 0.07, F change = 6.44, P < 0.05). Overall the results convincingly demonstrate that virtually all the first-order effects (from Table 1) can be accounted for by the length of the surgery, since when this variable is entered into a regression equation the other effects disappear. There are only two exceptions to this, both of which were found with GP2.
df
CI al.
Actions: (a) No follow-up Follow-up Referrals Investigations
No. of consultations GPI GP2 GP3 -0.55** 0.47’ 0.17 0.35
-0.13 0.52** -0.05 -0.42’
-0.44. 0.27 0.36 0.2 I
Nota: (a) Proportion of consultations with the action during the course of a surgery. GPl n = 20, GP2 n = 23, GP3 n = 21. lf < 0.05, lP < 0.01, l**P < 0.001.
Coping strategies
The data were further analysed to see whether the GPs employed specific coping strategies when faced with surgeries with a large number of consultations; that is, did the relative proportions of certain types of actions resulting from a consultation change with busier surgeries (see Table 3)? To look at this, the frequency of each type of action during the course of a surgery was divided by the total number of consultations in that surgery to produce a proportion score. This was then correlated with the number of consultations in a surgery. As can be seen from Table 3, a number of significant correlations emerged. These showed that as surgeries became busier, the proportion of consultations when follow-ups were arranged increased (for GPs 1 and 2), whereas the proportion of consultations when follow-ups were not arranged decreased (for GPs 1 and 3). However, contrary to this pattern GP2 was also found to be less likely to request investigations during busy surgeries. DISCUS!SlON The main aim of the present study was to identify some of the determinants of GPs’ subjective experience of surgery workload. It was hypothesised that not only would these be related to the actual amount of surgery workload (e.g. surgery length), but also to the type of work involved (e.g. r-e-presentations) and to the GPs’ perceptions of their consulting patients (e.g. the ‘need’ for the consultation). Correlational analyses revealed the main predictors of the GPs’ subjective workload ratings to centre around the objective surgery work demands, in terms of both the amount of work (i.e. surgery length and over-run, number of consultations) and the type of work conducted (i.e. number of re-presentations, (no)follow-ups and investigations requested). In contrast the GPs’ perceptions of their patients’ ‘need’ for each consultation were found to be largely unrelated to their subjective experience of surgery workload. This finding goes against previous work which suggests GPs believe many of their surgeries to be made up of trivial, unnecessary, and inappropriate consultations and that this might lead to feelings of stress [2]. Similarly, little evidence was found to link the GPs’ perceptions of their patients’ health status, which might heighten feelings of responsibility [15], with their subjective workload ratings. A series of regression analyses, in which surgery length was entered first into the equation, was
GPs’ subjective experience of their workload
employed on a reduced set of predictor variables in order to identify the best determinants of the GPs’ subjective experience of surgery workload. In this way it was possible to assess the influence of the other predictor variables over and above that accounted for by the length of a surgery. The results of these analyses convincingly demonstrated that virtually all of the first-order findings could be accounted for by surgery length, thus confirming the hypothesised link between objective surgery work demands and the GPs’ subjective experience of their workload. This is important as it provides direct confirmation of a previously assumed relationship. It is also important methodologically as it highlights the need to control for surgery length when assessing the importance of other factors. So for instance, when considering the influence of a particular type of consultation on subjective surgery workoad ratings, it has to be remembered that such consultations are likely to occur more frequently in longer surgeries. If this is not controlled for then it is not clear whether any effect is due to this type of consultation or the sheer overall bulk of consultations conducted in longer surgeries. The regression analyses showed that surgery length alone could account for between 52% and 66% of the variance in the GPs’ ratings of subjective workload. Only two variables were found to add significantly to the amount of variance explained. However, these appeared with only one of the GPs. So, for GP2 the number of investigations requested during the course of a surgery was also found to be related to subjective workload ratings. This was a negative relationship such that surgeries with many requests for investigations were given low ratings for subjective workload. The implication is that for this GP, requesting investigations may be a means of reducing the subjective level of his workload. In addition, the number of consultations in a surgery was also found to add to the amount of variance explained. This suggests that, for this GP, it may be the actual number of patients seen over the course of a surgery, on top of the time spent with them, that is leading to high subjective workload ratings. As was argued earlier, when faced with busy surgeries, GPs may employ certain coping strategies. Some evidence for this was found in the present study such that during long surgeries the GPs tended to arrange more follow-up appointments, whereas during shorter surgeries they tended against arranging follow-up appointments. However, such a strategy in response to a busy surgery may simply store up work for a future date [ 17] and, as the present findings suggest, its employment may be, at least in part, a response to time management rather than a purely clinical decision. To conclude, the small number of GPs involved in the study and the relatively short time scale employed mean that we should beware of making too strong a conclusion from the data; but they do provide a useful first insight into some of the factors which may influence GPs’ perceptions of their workload and the ways in which they may react to it. In particular, the correlational analyses went some way to contirm the link between surgery work demands, in both amount and type, and GPs’ subjective
165
experience of their workload. Such a link was not found with the GPs’ perceptions of their consulting patients. However, regression analyses revealed subjective workload ratings to be mainly accounted for by surgery length. This is not to say that many of the variables considered in the present study are not important work-related concerns for GPs and that they may make the GP’s job inherently stressful, only that they do not appear to predict variations in GPs’ reactions to their workload over and above that explained by the length of surgeries. Moreover, the finding that GPs in the present study tended to arrange more follow-up appointments during busy surgeries suggests that future work may consider in more detail the ways in which GPs try to cope with high work demands and their psychological reactions to them. REFERENCES
Bates E. Doctors and their spouses speak: stress in medical practice. Sot. Sri. Med. 20, 659-666. 1982. Cartwripht A. and Anderson R. General Practice Reoisited. Tavistock, London, 1981. Monice J. K. W. Job stress and burnout. Bull. R. Coil. Psychiatrists 8, 45-46, 1984. Cooper C. L., Rout U. and Faragher B. Mental health, job satisfaction, and job stress-among general practitioners. Br. Med. J. 298. 366-370. 1989. Registrar General Decennial Supplement for England and Wales, HMSO, London, 1978. Murray R. M. Alcoholism amongst male doctors in Scotland. Lnncet iii, 729-733. 1976. Fleming D. M. Workload review. J. R. Coll. Gen. Practioners 32. 292-297. 1982. Fry J. and Dillane J. B. Workload in a general practice 1950-1985. J. R. CON. Gen. Practioners 36, 403-404, 1986. 9. Mills K. A. and Reilly P. M. General practitioner workload with 2,000 patients. The Ulster Med. J. 55, 334, 1986. 10. Department of Health and Social Security General Medical Practitioners’ Workoad, HMSO, London. 1987. 11. Bain J. How hard do general practitioners work? Br. Med. J. 289. 14741475. 1984. 12. Bosenquet N. Fresh light shed on GPs’ workload. H&h Service 1. 2, 753, 1987. 13. Camey T. Workload of general practitioners. Br. Med. J. 299, 753, 1989. 14. Lowe M. Are doctors working harder? BMA News Rev. 11, 12-13, 1985. IS. Kaltz D. and Kahn R. L. The Social Psychofogy of Organizations, 2nd edn. Wiley, New York, 1978. 16. Mechanic D. Politics, Medicine and Social Science. Wiley, New York, 1974. 17. Porter A. M. D.. Howie J. G. R. and Levinson A. Stress and the general’ practitioner. In Stress ia Heafrh Professionals (Edited by Payne R. and Fir&Cozens J.). Wiley, London, 1987. 18. Morrell D. C. and Nicholson S. Measuring the results of changes in the method of delivering primary medical care-a cautionary tale. J. R. CON. Gen. Pracfitioners 24, 111-118, 1974. 19. Richardson I. M., Howie J. G. R.. Dumo D.. Gill G. and Dingwall-Fordyce I. A study generalpractitioner consultations in North-East Scotland. J. R. Co/l. Gen. Practitioners 23, 132-142. 1973. 20. Rosser R. M. and Watts V. C. The measurement of hospital output. In!. J. Epidemiol. 1, 361-368, 1972.
of
166
PAUL NOIUAN et al.
21. Hart S. G. and Staveland L. E. Development of NASA-TLX (Task Load Index): results of empirical and theoretical research. In Human Mental Workload (Edited by Hancock P. A. and Meshkati N.). Elsevier, Amsterdam, 1988. 22. Sheridan T. B. and Simpson R. W. Toward the Definition and Measurement of the Mental Workload of Transport Pilots. Massachusetts Institute of Technology Flight Transportation Laboratory Report, Cambridge, MA, 1979.
23. Box G. E. P. and Jenkins G. M. Time Series Analysis: Forecasting and Control. Holdcn Day, San Francisco, 1976. 24. Box G. E. P., Hunter W. G. and Hunter J. S. Stutiptics for Experimenters. Wiley, New York, 1978. 25. McCleary R. and Hay R. A. Applied Time Series Analysis for the Social Sciences. Sage, Beverley Hills, 1980. 26. Cohen J. and Cohen P. Applied Mulriple Regresston/ Correlation Analysis for the Behavioral Sciences, 2nd edn. Lawrence Erlbaum Associates, Hillsdale, NJ, 1983.
GENERAL
Copyright
PRACTITIONERS’ OF SURGERY
SUBJECTIVE WORKLOAD
0
0277-9536/91 S3.00 + 0.00 1991 Pergamon Press pit
EXPERIENCE
PAUL NORMAN,MIKE F~I-IERand TOBY WALL MRC/ESRC Social and Applied Psychology Unit, Department of Psychology, University of Sheffield, Sheffield SIO 2TN, England Abstract-General practitioners (GPs) in the U.K. frequently cite heavy workload as a major source of work-related stress. However, there is little direct evidence either to support this claim or to suggest whether the source of this stress lies in the amount, or rather the type, of work undertaken. The present paper uses a within-subjects design to address these issues in the surgery context. Over a two-week period three GPs gave ratings of subjective workload following each of their surgeries (N = 20-23). Correlational analysis showed a wide range of surgery characteristics to predict the GPs’ nactions, including the length of surgeries, the number of consultations, the number of re-presentations, and the number of follow-ups and investigations requested. The GPs’ perceptions of their patients (e.g. in terms of their health status and the appropriateness of the consultation), in contrast, did not emerge as significant predictors. The results of multiple regression analyses, however, revealed that most of the first-order effects could be accounted for by surgery length. Thus while the study supported the hypothesised link between the overall level of surgery work demands and the subjective experience of workload, it failed to confirm any consistent additional effects of the type of work involved. It was also found that as the length of surgeries increased, the GPs were more likely to arrange follow-up appointments. The implications of the present findings are therefore two-fold. First, that future research on surgery work demands and the subjective experience of workload must take account of surgery length as a primary predictor; and second, that attention be paid to how GPs cope with heavy workloads. Key words-general
practitioners,
surgery workload, subjective experience, coping strategies
INTRODUCTION
Many general practitioners (GPs) in the U.K. consider themselves to be under stress at work, and point to the amount and nature of their workload as major contributing factors [l]. Three strands of evidence give credence to this claim. First, surveys on GPs’ attitudes to their work have shown that many report a feeling of ‘lack of time’ and that they find their work to be frustrating and stressful [2,3]. For example, in a sample of Aberdeen GPs, 87% reported having experienced periods of exhaustion’ (31. Second, the demands of the GP’s work have been linked to low levels of job satisfaction and psychological well-being [4]. Third, there is objective data to suggest that GPs may be at particular risk of illnesses and other behavioural indicators of stress [5]. For instance, in Scotland, the rate of first admissions for alcohol dependence has been found to be two to seven times higher amongst GPs than comparable controls [6]. Such evidence has prompted other investigators to map the objective work demands faced by GPs [7-g]. A study conducted by the DHSS [lo] of over 1000 GPs found the average GP to be someone who spends almost half his/her time each week conducting nearly 150 surgery consultations. Whilst such descriptive data has fuelled a continuing debate surrounding the acceptability of GPs’ work demands [ll-141, it fails to demonstrate a direct link between heavy work demands and potential negative reactions of the kind which are associated with stress at work. To know that the average GP conducts nearly 150 surgery consultations per week is not informative in itself. It is unclear whether this should be considered to be
high or low, stressful or not. To address the question more directly a minimum requirement would be to show that as the number of consultations vary so too do GPs’ psychological reactions. Implicit in much of the literature on GP’s work demands is the view that it is not only the amount of work that GPs undertake that is stressful, but also the nature of that work. Job stress research generally has shown ‘responsibility for others’ to be a key workrelated stressor [IS]. So, it may the case that when GPs are faced with patients who are under great physical or emotional distress, their awareness of this responsibility is heightened and this may be a source of stress. On the other hand though, there is evidence to suggest that GPs find many of their consultations to be either trivial, unnecessary, or inappropriate and that this leads to a feeling of frustration [2]. This may in turn affect the quality of job performance [16]. Either way, it is likely that GPs’ reactions to their consulting patients will be related to their perceptions of, and psychological reactions to, their surgery workload. For such factors to be taken seriously as potential work-related stressors for GPs, it needs to be shown that GPs’ reactions to their work change as a function of these factors. When faced with heavy surgery work demands, GPs may employ certain coping strategies. For instance it has been suggested that the cutting short of consultations by asking patients back at a later date may be a strategy which is frequently employed by GPs when faced with busy surgeries [17]. In fact, it has been documented that about 60% of consultations are GP-initiated re-presentations [18, 191. However, it may be the case that this response simply
161
PAUL NORMANer al.
162
stores up the work for the future, and that it will be the cause of additional stress when the patients re-present [17]. Thus, this type of consultation may be a particular work-related stressor for GPs. Given that GPs frequently report heavy workload as a stressor, that there is evidence to link this to morbidity and mortality statistics, and that they spend almost half their work-time in surgery consultations, it was decided to conduct a study to explore the determinants of their subjective experience of surgery workload. This could be achieved in one of two main ways. The first is to study the experiences of GPs working under different levels or types of work demand, using an independent subjects research design. This has the advantage of allowing larger scale investigations but, because of the difficulty of holding third factors constant (e.g. practice characteristics) it has interpretational difficulties. The second approach is to adopt a within subjects, or time series, research design in which variations in work demands over time are related to psychological reactions for the same GPs working under otherwise equivalent circumstances. Both approaches are required and complement each other. However, it is the time series approach which is adopted here, where we investigate the relationship of both the amount and type of surgery work undertaken by three GPs to their subjective experience of surgery workload. More specifically, on the basis of previous research, a number of hypotheses were formulated: firstly, that high subjective workload ratings would be related to the number of consultations in a surgery and its length (i.e. the amount of work); secondly, that variations in certain types of consultations (e.g. re-presentations) would be associated with the GPs’ subjective workload ratings over and above that accounted for by the total number of consultations in a surgery (i.e. the type of work); and thirdly, that the GPs’ perceptions of their consulting patients (e.g. of their emotional and physical state, and of the appropriateness of the consultation) would also influence ratings of subjective workload. Finally, it was hypothesised that the GPs may employ certain coping strategies (e.g. referring and arranging follow-up appointments) when faced with busy surgeries. METHOD
The study focus
The study was conducted at a single practice consisting of three male GPs with a combined list size of approximately 7000 patients. The practice is situated in a town in the East Midlands, England. The three GPs completed short questionnaires at the end of each consultation and each surgery over a twoweek period. During this period 733 consultations were conducted in 68 surgeries. Questionnaires were returned for 688 of the consultations (93.9% response rate) and for 67 of the surgeries (98.5% response rate). Surgery characteristics
The three GPs provided detailed data on the characteristics of each surgery they conducted over the two-week study period. Firstly, they recorded the
type of each consultation. These fell into two main categories: presentations and re-presentations. Secondly, they recorded any actions resulting from the consultation. These fell into four main categories: no follow-up arranged, follow-up with the GP arranged, referral, and the requesting of investigations. For further analysis these data were summed for each surgery so that the influence of the make-up of surgeries (in terms of the number of different types of consultations and resulting actions) on the subjective experience of surgery workload could be assessed. In addition, the GPs assessed each patient’s health status. Thus a health status scale was produced for each patient by combining the GPs’ ratings along three dimensions: disability (7 point scale: no disability, slight social, affects work, cannot work, needs help, bed-bound, unconscious), emotional distress and pain/physical distress (4 point scales: none, mild, moderate, severe). These measures were based on scales developed previously [20]. Finally, the GPs rated on 5 point scales (very high, high, medium, low, very low) their perceptions of each patients ‘need’ for the consultation. In short, they rated the urgency (“how urgent was the patient’s need for the consultation?‘), the importance (“how important was it that the patient attend the consultation in terms of an ongoing state of health?‘), the appropriateness (“how appropriate was the consultation in terms of your specific role in the Primary Health Care Team?‘), and the helpfulness (“how helpful do you feel the consultation was for the patient?“) of each consultation. Again, these ratings were combined to produce a consultation appropriateness score. For analysis the above ratings were averaged for each surgery in order to examine whether the GPs’ perceptions of their consultations over the course of a surgery affected their subjective experience of surgery workload. The GPs also indicated the start and end time of each surgery. From this it was possible to calculate the length of each surgery. Surgery workload questionnaire
The questionnaire employed in the present study to assess each GP’s subjective experience of their workload was a slightly modified version of the NASA Task Load Index, which has been developed, tested and used in a range of experimental situations to measure work-related stress [21]. The three GPs therefore rated each surgery in terms of several perceived aspects of their workload [22]. These were termed mental demand [“How much mental and perceptual activity (e.g. searching for information, making decisions, performing calculations) was required? Were your duties easy or demanding?“]; physical demand (“How much physical activity was required? Was the physical demand slack or strenuous?‘); temporal demand (“How much time pressure did you feel under as the result of the rate of consultations? Was the pace slow and leisurely or rapid and frantic?” ); performance (“How successful do you think you were in accomplishing the goals of your duties?‘); efort (“How hard did you have to work (mentally and physically) to accomplish your level of performance?‘); andfncstration (“How much frustration did you feel during the surgery? Were YOU content and relaxed or discouraged and irritated?“).
GPs’
subjective experience of their workload
These reactions were assessed via ten-point rating scales (low/high anchors) covering each of the six aspects, with a 0 therefore indicating low perceived demand and a 9 high perceived demand. Initial analyses revealed scores on these six items to be highly intercorrelated. Combining them to produce an overall workload scale for each GP yielded reasonably high alpha coefficients (see Table I), thus indicating satisfactory levels of internal consistency. Furthermore, the scale was found to relate to the predictor variables in the same way as the individual component items. For the following analyses overall workload scale scores have therefore been used. Data analysis
The data from each GP were analysed as a separate time series, following procedures described by Box and Jenkins [23], and others [24,25]. Since repeated measurement using the same instrument could influence the scores obtained (either through recall of previous responses or through ‘practice’ effects), the first stage of analysis was to examine each time series for the presence of serial dependences and trends. No such effects were found, with the exception of a small and isolated auto-correlation (lag = 1) effect for one GP only. Consequently, there was no need to take account of such effects in subsequent analyses. In addition, given the limited size of the data set, each variable was examined to determine whether it displayed sufficient variation to warrant further investigation. This was found to be the case for all the variables considered for each of the GPs. Analysis then proceeded in two further stages. First, the relationship between each of the independent variables and the dependent variable of subjective workload was tested by calculating product-moment correlations from each GP’s data, thus testing for first-order effects. However, since several of the independent variables were logically and empirically interrelated (e.g. longer surgeries tend to have more consultations, more referrals, etc.), the second stage involved multiple regression analyses. Here, for each GP, length of surgery was entered first into the regression equation on the basis that conceptually and empirically it was the primary predictor of the subjective experience of surgery workload, followed by any other variables explaining additional variance. This therefore tested for the effect of these other surgery characteristics over and above that accounted for by surgery length. However, given the close interdependence of many of the variables considered here, as highlighted above, there is a potential problem of multicollinearity. Since this has implications for the validity of regression techniques by making beta weights unstable [16], it was important to determine its relevance in the present case. To address this two steps were taken. First, correlation matrices of the independent variables for each GP were computed and analysed to determine whether any variables were interrelated to an extent that might invalidate regression analyses. This was not found to be the case. Second, in conducting the regression analyses, SPSSx was employed which calculates the tolerance levels of the independent variables and therefore provides a test SSM 3312-E
163
for whether multicollinearity is a problem. Again, this was not found to be the case. RESULTS
First -order effects
The results of the first-order analyses are presented in Table 1. Five variables were found to stand out as strong predictors of the subjective experience of surgery workload for all three GPs. These were the length of the surgery, the overall number of consultations, the number of re-presentations, the number of consultations when no follow-up was arranged, and the number of requests made for investigations during the course of the surgery. In addition, the number of follow-ups arranged during the course of the surgery predicted the ratings for two of the three GPs. It is evident therefore that the main predictors of the GPs’ subjective workload ratings centred around the objective surgery work demands, both in terms of the amount of work conducted (e.g. surgery length) and its type (e.g. re-presentations). By contrast, the GPs’ perceptions of their consulting patients, in terms of their health status and their ‘need’ for the consultation, were largely unrelated to their subjective workload ratings. In parentheses: it is worth noting that given the centrality of surgery length (or time) as a predictor of the subjective experience of surgery workload, we considered the predictive value of alternative variants of this core construct. Following the suggestions of an anonymous referee, two more variables were created. Both centred on the amount of work involved in relation to time. The first was a ratio score of time in relation to numbers seen and the second was a measure of the extent to which the surgeries over-ran. The over-run measure correlated highly with workload ratings for all three GPs (rs = 0.84, 0.79, 0.67, Ps < 0.001) as did the ratio score for GPl and GP3 (rs = 0.64, 0.51, Ps c 0.01.0.05). However, given that neither of these measures explained variance in addition to surgery length, and being derived directly from this core variable introduce problems Table 1. Zero order correlates of subjective workload ratings for each GP GP: Amount of work Surgery length Number of consultations Type of work Presentations Re-presentations Actions from consultations: No follow-up Follow-up Referrals Investigations Perceptions: Patient health status Consultation appropriateness Mean: SD: Alpha:
Subjective workload ratings I 2 3 0.81*** 0.68’9.
0.77*** 0.76’**
0.72*** 0.54**
0.67*** 0.40.
0.21 0.66***
0.29 0.67***
0.46’ 0.60.. -0.07 0.40.
0.44. 0.53** 0.26 -o.so**
-0.04 0.26
0.32 -0.04
0.35 0.37.
31.70 9.00 0.93
36.39 6.63 0.73
40.19 7.18 0.76
Notes: GPI n = 20, GP2 n = 23, GP3 n = 21. lf c 0.05, l*p < 0.01, l**p < 0.001.
0&O** -0.07 0.08 0.43.
164
PAULNOW
of multicollinearity, they were not considered subsequent analyses. Second-order
in
Table 2. Regression analyses: significant independent predictors of each GP’s subjective experience of their surgery workload (beta values1 GP:
Subjective workload ratings I 2 3
s1ep 1: &Ill
vahes Surgery length
F-value df R Square Stev 2: Beta values Surgery length Investigations F-value
4 R Square
o.s1***
3J.20*** (1.18) 0.66
0.779.. 30.17999 (1.21) 0.59
0.72+** 20.26**’ (ii;)
.
0.709’. -0.36” 25.12*** (2.20) 0.72
Step 3: Beta calues
Surgery length Investigations Number of consultations F-value R Sauare
0.41. -0.34. 0.40. 23.45*** (3.19) 0.79
Notes: GPI n = 20, GP2 n = 23, GP3 n = 21. lP < 0.05, l*P c 0.01. ***F- < 0.001.
Tabk 3. Correlations between the number of consultations in a surgery and the employment of different actions
effects
The second step in identifying the best predictors of each GP’s subjective experience of their surgery workload was to enter those variables which were found to significantly correlate for all three GPs into a series of stepwise regression analyses. Thus for each GP the five predictor variables outlined above were used to predict their ratings of subjective workload, but with surgery length being the tirst variable to be entered into the regression equation. This is clearly essential because, if the number of consultations, re-presentations, follow-ups not arranged, or investigations requested over the course of a surgery are to be taken seriously as potential stressors, it must be demonstrated that their effects exist over and above the fact that they simply occur more frequently in longer surgeries. The results of the regression analyses are presented in Table 2, and point to two main findings. Firstly, surgery length was found to be a significant predictor for each of the three GPs when entered by itself, accounting for between 52% and 66% of the variance in the subjective workload ratings. Secondly, when the other predictor variables were entered into the regression equations only two significant predictors emerged; both for GP2, for whom the number of investigations requested over the course of a surgery accounted for a significant proportion of the variance in subjective workload ratings over and above the effect of surgery length (R square change = 0.13, F change = 8.83, P < O.Ol), as did the overall number of consultations (R square change = 0.07, F change = 6.44, P < 0.05). Overall the results convincingly demonstrate that virtually all the first-order effects (from Table 1) can be accounted for by the length of the surgery, since when this variable is entered into a regression equation the other effects disappear. There are only two exceptions to this, both of which were found with GP2.
df
CI al.
Actions: (a) No follow-up Follow-up Referrals Investigations
No. of consultations GPI GP2 GP3 -0.55** 0.47’ 0.17 0.35
-0.13 0.52** -0.05 -0.42’
-0.44. 0.27 0.36 0.2 I
Nota: (a) Proportion of consultations with the action during the course of a surgery. GPl n = 20, GP2 n = 23, GP3 n = 21. lf < 0.05, lP < 0.01, l**P < 0.001.
Coping strategies
The data were further analysed to see whether the GPs employed specific coping strategies when faced with surgeries with a large number of consultations; that is, did the relative proportions of certain types of actions resulting from a consultation change with busier surgeries (see Table 3)? To look at this, the frequency of each type of action during the course of a surgery was divided by the total number of consultations in that surgery to produce a proportion score. This was then correlated with the number of consultations in a surgery. As can be seen from Table 3, a number of significant correlations emerged. These showed that as surgeries became busier, the proportion of consultations when follow-ups were arranged increased (for GPs 1 and 2), whereas the proportion of consultations when follow-ups were not arranged decreased (for GPs 1 and 3). However, contrary to this pattern GP2 was also found to be less likely to request investigations during busy surgeries. DISCUS!SlON The main aim of the present study was to identify some of the determinants of GPs’ subjective experience of surgery workload. It was hypothesised that not only would these be related to the actual amount of surgery workload (e.g. surgery length), but also to the type of work involved (e.g. r-e-presentations) and to the GPs’ perceptions of their consulting patients (e.g. the ‘need’ for the consultation). Correlational analyses revealed the main predictors of the GPs’ subjective workload ratings to centre around the objective surgery work demands, in terms of both the amount of work (i.e. surgery length and over-run, number of consultations) and the type of work conducted (i.e. number of re-presentations, (no)follow-ups and investigations requested). In contrast the GPs’ perceptions of their patients’ ‘need’ for each consultation were found to be largely unrelated to their subjective experience of surgery workload. This finding goes against previous work which suggests GPs believe many of their surgeries to be made up of trivial, unnecessary, and inappropriate consultations and that this might lead to feelings of stress [2]. Similarly, little evidence was found to link the GPs’ perceptions of their patients’ health status, which might heighten feelings of responsibility [15], with their subjective workload ratings. A series of regression analyses, in which surgery length was entered first into the equation, was
GPs’ subjective experience of their workload
employed on a reduced set of predictor variables in order to identify the best determinants of the GPs’ subjective experience of surgery workload. In this way it was possible to assess the influence of the other predictor variables over and above that accounted for by the length of a surgery. The results of these analyses convincingly demonstrated that virtually all of the first-order findings could be accounted for by surgery length, thus confirming the hypothesised link between objective surgery work demands and the GPs’ subjective experience of their workload. This is important as it provides direct confirmation of a previously assumed relationship. It is also important methodologically as it highlights the need to control for surgery length when assessing the importance of other factors. So for instance, when considering the influence of a particular type of consultation on subjective surgery workoad ratings, it has to be remembered that such consultations are likely to occur more frequently in longer surgeries. If this is not controlled for then it is not clear whether any effect is due to this type of consultation or the sheer overall bulk of consultations conducted in longer surgeries. The regression analyses showed that surgery length alone could account for between 52% and 66% of the variance in the GPs’ ratings of subjective workload. Only two variables were found to add significantly to the amount of variance explained. However, these appeared with only one of the GPs. So, for GP2 the number of investigations requested during the course of a surgery was also found to be related to subjective workload ratings. This was a negative relationship such that surgeries with many requests for investigations were given low ratings for subjective workload. The implication is that for this GP, requesting investigations may be a means of reducing the subjective level of his workload. In addition, the number of consultations in a surgery was also found to add to the amount of variance explained. This suggests that, for this GP, it may be the actual number of patients seen over the course of a surgery, on top of the time spent with them, that is leading to high subjective workload ratings. As was argued earlier, when faced with busy surgeries, GPs may employ certain coping strategies. Some evidence for this was found in the present study such that during long surgeries the GPs tended to arrange more follow-up appointments, whereas during shorter surgeries they tended against arranging follow-up appointments. However, such a strategy in response to a busy surgery may simply store up work for a future date [ 17] and, as the present findings suggest, its employment may be, at least in part, a response to time management rather than a purely clinical decision. To conclude, the small number of GPs involved in the study and the relatively short time scale employed mean that we should beware of making too strong a conclusion from the data; but they do provide a useful first insight into some of the factors which may influence GPs’ perceptions of their workload and the ways in which they may react to it. In particular, the correlational analyses went some way to contirm the link between surgery work demands, in both amount and type, and GPs’ subjective
165
experience of their workload. Such a link was not found with the GPs’ perceptions of their consulting patients. However, regression analyses revealed subjective workload ratings to be mainly accounted for by surgery length. This is not to say that many of the variables considered in the present study are not important work-related concerns for GPs and that they may make the GP’s job inherently stressful, only that they do not appear to predict variations in GPs’ reactions to their workload over and above that explained by the length of surgeries. Moreover, the finding that GPs in the present study tended to arrange more follow-up appointments during busy surgeries suggests that future work may consider in more detail the ways in which GPs try to cope with high work demands and their psychological reactions to them. REFERENCES
Bates E. Doctors and their spouses speak: stress in medical practice. Sot. Sri. Med. 20, 659-666. 1982. Cartwripht A. and Anderson R. General Practice Reoisited. Tavistock, London, 1981. Monice J. K. W. Job stress and burnout. Bull. R. Coil. Psychiatrists 8, 45-46, 1984. Cooper C. L., Rout U. and Faragher B. Mental health, job satisfaction, and job stress-among general practitioners. Br. Med. J. 298. 366-370. 1989. Registrar General Decennial Supplement for England and Wales, HMSO, London, 1978. Murray R. M. Alcoholism amongst male doctors in Scotland. Lnncet iii, 729-733. 1976. Fleming D. M. Workload review. J. R. Coll. Gen. Practioners 32. 292-297. 1982. Fry J. and Dillane J. B. Workload in a general practice 1950-1985. J. R. CON. Gen. Practioners 36, 403-404, 1986. 9. Mills K. A. and Reilly P. M. General practitioner workload with 2,000 patients. The Ulster Med. J. 55, 334, 1986. 10. Department of Health and Social Security General Medical Practitioners’ Workoad, HMSO, London. 1987. 11. Bain J. How hard do general practitioners work? Br. Med. J. 289. 14741475. 1984. 12. Bosenquet N. Fresh light shed on GPs’ workload. H&h Service 1. 2, 753, 1987. 13. Camey T. Workload of general practitioners. Br. Med. J. 299, 753, 1989. 14. Lowe M. Are doctors working harder? BMA News Rev. 11, 12-13, 1985. IS. Kaltz D. and Kahn R. L. The Social Psychofogy of Organizations, 2nd edn. Wiley, New York, 1978. 16. Mechanic D. Politics, Medicine and Social Science. Wiley, New York, 1974. 17. Porter A. M. D.. Howie J. G. R. and Levinson A. Stress and the general’ practitioner. In Stress ia Heafrh Professionals (Edited by Payne R. and Fir&Cozens J.). Wiley, London, 1987. 18. Morrell D. C. and Nicholson S. Measuring the results of changes in the method of delivering primary medical care-a cautionary tale. J. R. CON. Gen. Pracfitioners 24, 111-118, 1974. 19. Richardson I. M., Howie J. G. R.. Dumo D.. Gill G. and Dingwall-Fordyce I. A study generalpractitioner consultations in North-East Scotland. J. R. Co/l. Gen. Practitioners 23, 132-142. 1973. 20. Rosser R. M. and Watts V. C. The measurement of hospital output. In!. J. Epidemiol. 1, 361-368, 1972.
of
166
PAUL NOIUAN et al.
21. Hart S. G. and Staveland L. E. Development of NASA-TLX (Task Load Index): results of empirical and theoretical research. In Human Mental Workload (Edited by Hancock P. A. and Meshkati N.). Elsevier, Amsterdam, 1988. 22. Sheridan T. B. and Simpson R. W. Toward the Definition and Measurement of the Mental Workload of Transport Pilots. Massachusetts Institute of Technology Flight Transportation Laboratory Report, Cambridge, MA, 1979.
23. Box G. E. P. and Jenkins G. M. Time Series Analysis: Forecasting and Control. Holdcn Day, San Francisco, 1976. 24. Box G. E. P., Hunter W. G. and Hunter J. S. Stutiptics for Experimenters. Wiley, New York, 1978. 25. McCleary R. and Hay R. A. Applied Time Series Analysis for the Social Sciences. Sage, Beverley Hills, 1980. 26. Cohen J. and Cohen P. Applied Mulriple Regresston/ Correlation Analysis for the Behavioral Sciences, 2nd edn. Lawrence Erlbaum Associates, Hillsdale, NJ, 1983.