Divergent Intentions to Use Antibiotic Guidelines

Abstract

Background. To improve physicians’ antimicrobial practice, it is important to identify barriers to and facilitators of guideline adherence and assess their relative importance. The theory of planned behavior permits such assessment and has been previously used for evaluating antibiotic use. According to this theory, guideline use is fueled by 3 factors: attitude, subjective norm (perceived social pressure regarding guidelines), and perceived behavioral control (PBC; perceived ability to follow the guideline). The authors aim to explore factors affecting guideline use in their hospital. Methods. Starting from their earlier observations, the authors constructed a questionnaire based on the theory of planned behavior, with an additional measure of habit strength. After pilot testing, the survey was distributed among physicians in a major teaching hospital. Results. Of 393 contacted physicians, 195 completed questionnaires were received (50.5% corrected response rate). Using multivariate analysis, the overall intention toward using antibiotic guidelines was not very predictable (model R² = .134). Habit strength (relative weight = .391) and PBC (relative weight = .354) were the principal significant predictors. A moderator effect of respondents’ position (staff member v. resident) was found, with staff members’ intention being significantly influenced only by habit strength and residents’ intention by PBC. Regarding previously identified barriers, education on antibiotics and guidelines was rated unsatisfactory. Conclusions. These divergent origins of influence on guideline adherence point to different approaches for improvement. As habits strongly influence staff members, methods that focus on changing habits (e.g., automated decision support systems) are possible interventions. As residents’ intention seems to be guided mainly by external influences and experienced control, this may make feedback, convenient guideline formats, and guideline familiarization more suitable.

Keywords

guidelines antibiotics questionnaire theory of planned behavior

Introducing and maintaining guidelines—and, more specifically, local guidelines—are key instruments in the rational use of antibiotics.¹ Their positive effects on antimicrobial use, costs, and therapeutic outcome have been confirmed several times.^2–5 Notwithstanding these advantages, guideline implementation is hindered by a number of barriers, and their actual use remains erratic.⁶

Techniques such as interviews or focus groups are available to discern possible barriers,^7–9 but a supporting theory is needed to situate the results, allowing for targeted interventions and improved efficiency.¹⁰ One often used theory, the theory of planned behavior (TPB), is developed to describe the determinants of individual intentions and behaviors—in this case, guideline use. Basically, an intention regarding a certain behavior is a function of 3 determinants: 1) attitude toward the behavior, referring to the positive or negative evaluation of performing the behavior; 2) subjective norm, reflecting the person’s perception of social pressure regarding the performance; and 3) perceived behavioral control, which refers to the person’s perception of own control over performance of the behavior.¹¹ Subsequent intention will then lead to the final behavior. This theory has already been successfully used several times in the evaluation of medical practice, including the field of antibiotic usage.^12–15

The overall goal of our study was to enhance the understanding of how guidelines are used in our hospital and how this can be improved. Therefore, we constructed and distributed a questionnaire based on TPB, adapted to cover possible habits. On the basis of previous work in this area, we expected that past behavior and thus also habits can add to the prediction of behavioral intention and, through intention, of actual behavior.^14,16 With this survey, we wanted further to assess the impact of barriers against our local guideline that were revealed in our earlier focus group study and possible differences between physicians’ groups.⁸

Methods

This study is part of a larger project on antibiotic use in the hospital environment (ClinicalTrials.gov number: NCT00512772) and is approved by the Ethics Committee of University Hospitals Leuven, Belgium. It was conducted in a large (1900-bed) tertiary-care, university teaching hospital. The studied guidelines are the locally developed antibiotic guidelines, available as a booklet, on the hospital intranet and the Internet (http://www.antibioticagids.be) with a last revision, before the survey, in June 2007.

Questionnaire Development

Our questionnaire consisted of measures of the main components of TPB (i.e., attitudes, subjective norm, and perceived behavioral control), habit, and the intention toward using antibiotic guidelines. The overall questionnaire was constructed according to recommendations,^11,17 comprising both direct and indirect measures (see below) and using the validated Dutch translation.¹⁷ The measures that were used are described in more detail below.

Intention (INT)

Three items were used to measure the intention toward using antibiotic guidelines (“During the next year, I expect to/want to/intend to use the guidelines for my patients with a suspected infection”). Items were scored on a 7-point (1 = strongly disagree to 7 = strongly agree) Likert scale.

Attitude (ATT)

We assessed attitudes toward using antibiotic guidelines with 6 items. The question read, “Using antibiotic guidelines during the next year for my patients with a suspected infection is . . .,” followed by 6 items with bipolar labels on each end of the 7-point scale: harmful–beneficial, useless–useful, bad–good, annoying (for me)–pleasant (for me), not giving (me) satisfaction–giving (me) satisfaction, and worthless–valuable.

Subjective norm (SN)

Four items were used to measure subjective norms. Items read, “People who are important to me want me to use the guidelines,” “People who are important to me use the guidelines themselves,” “It is expected of me that I use the guidelines,” and “I feel social pressure to use the guidelines.” All items were scored on a 7-point (1 = strongly disagree to 7 = strongly agree) Likert scale.

Perceived behavioral control (PBC)

To measure PBC, the following 5 items were used: “The decision to use the guidelines is beyond my control” (reversed), “It is entirely up to me to use the local antibiotic guidelines,” “It is easy for me to use . . .,” “I feel capable to use . . .,” and “Using the guidelines is easy–difficult for me.” Items were scored on a 7-point Likert scale (1 = strongly disagree to 7 = strongly agree or 1 = easy to 7 = difficult).

Habit (HAB)

It has already been shown that past use of antibiotic guidelines (i.e., past behavior) is positively associated with the present intention to prescribe antibiotics according to guidelines.¹⁴ In that previous study, it was suggested that the explanatory power TPB can be improved by the addition of past behavior. Therefore, to evaluate the influence of habit on guideline use, we added the 12-item Self-Report Index of Habit Strength.¹⁶ Here, this concept of habit includes the automatic character of the behavior that is broader and also more precise than self-reported past behavior.¹⁸ All items of this construct were scored on a 7-point (1 = strongly disagree to 7 = strongly agree) Likert scale. This construct, similar to TPB constructs, loads on a single factor used in the final analysis.

Indirect Measures

In addition to the “direct” measures above, it is recommended to include so-called “indirect” measures for ATT, SN, and PBC.^17,19 This is based on the assumption that the direct measurement of these determinants using standardized constructs is also the reflection of a combination of actual beliefs and opinions. Indirect measures are tailored question sets that sufficiently cover the breadth of ATT, SN, or PBC in regard to the specific behavior. They should correlate well with the direct measure and can possibly point to barriers and facilitators with a high impact. Therefore, possible barriers against the use of antibiotic guidelines were identified by focus group discussions where all major groups (internal medicine and surgery, residents, and staff members) within this survey were represented. The results of this focus group study are described in detail elsewhere.⁸ Identified barriers were sorted according to whether they belonged to ATT, SN, or PBC by 2 independent researchers (P-JC and GL) with differences resolved by consensus. Within each category, the top 75% barriers with highest quotation frequency were rephrased into items and added to the questionnaire. For ATT, 6 supplementary indirect questions were included, 5 were added for SN, and 11 questions were added for PBC. Indirect items were scored on a 7-point Likert scale and formulated as recommended.¹⁷ For SN, moderating (“motivation to comply”) questions for each indirect item were included, and after recoding the indirect items to a −3 to +3 scale, their pairwise multiplication was used for further analysis. (For example, if the score on the item “the staff members in this hospital think I should use the antibiotic guidelines” was 6 [“agree”; 2 after recoding to a −3 to +3 scale] and the corresponding score on the motivation-to-comply question “what staff members think I should do is important to me” was 4 [the middle on the 1–7 scale of importance], the resulting product would be 2 × 4 = 8, on a −21 to +21 scale.) For ATT and PBC, such questions proved to have practically no variability among participants in the pilot study and were omitted in the final version of the questionnaire (see below).

Our focus group study demonstrated rather negative opinions concerning whether a microbiologist or a clinical pharmacist should provide additional antibiotic guidance, in order to relieve the infectious disease specialist.⁸ Therefore, 3 supplementary questions separately assessed on a 7-point (1 = fully disagree to 7 = fully agree) Likert scale the approval of an infectious disease specialist, a microbiologist, and a clinical pharmacist giving supplementary antimicrobial support. As a final check for additional items, the resulting survey was compared with other questionnaires.^13,20–22

Demographics

Participants were asked for their gender, age, hierarchical position (staff member v. resident), discipline (internal medicine, surgery, emergency medicine), and, as a measure of physicians’ general acquaintance with antimicrobials, how often they prescribed antibiotics on average per week (never, 1–2 times, 3–4 times, daily).

Pilot Study

This questionnaire was piloted in the pediatric department of the same hospital among 10 randomly selected physicians with various levels of experience and holding various positions. Participants were asked to complete the questionnaire with additional questions on readability, understandability, redundant or missing topics, possibly provocative topics, ease of use, and time needed to fill in the questionnaire. The pilot revealed no major remarks on the contents or form, with an acceptable fill-in time between 8 and 20 minutes.

Participants

All physicians, residents, and staff members working in our hospital at the time of the study were eligible for participation. Disciplines with intrinsically low antibiotic use or specific type of patients with limited guidelines available were excluded: anesthesiology, dermatology, ophthalmology, obstetrics–gynecology, radiology, pediatrics, psychiatry, and physicians working in laboratory departments. Also infectious diseases specialists were excluded due to their involvement in this project. The heads of department were contacted to gain explicit support for this study. The final number of eligible physicians was 393 for our hospital (214 residents, 179 staff members). The questionnaire (see Web appendix, unvalidated translation and Flemish-Dutch original) was distributed in 3 rounds: A first round was sent in the first half of February 2009; after 5 weeks, a reminder was sent; and finally after another 4 weeks, the entire questionnaire was resent to the remaining nonrespondents. As an incentive, a donation of €10 to the hospital care fund for children with cancer was made for each returned completed form.

Analysis

For each measure, both direct and indirect, the average was calculated and used in the analysis. To measure the internal consistency of the measures, Cronbach’s alpha was calculated. If alpha was lower than 0.70, items were removed to increase internal consistency.^17,23 To explore significant associations, we calculated Pearson’s correlations for all variables, except for gender and position (resident v. staff member) where point-biserial correlations were used. Mutual associations between gender, position, and discipline were given using Cramer’s V. Hierarchical linear regression analysis was performed with intention as the dependent variable. In the first step, sociodemographic variables (respondents’ hierarchical position, age, gender, and prescribing frequency) were entered to account for factors referred to as “external variables” in TPB. Position and gender were dummy coded. In the second step, the basic TPB variables (ATT, SN, PBC) were added to the equation. Finally, HAB was entered in the analysis as a supplementary independent variable as this concept is not included in TPB. Together with the R², the relative weight was calculated for each variable to assess their individual impact in the final model.²⁴ Indirect items were explored for interesting differences according to gender, position, or between surgery and internal medicine using the Mann-Whitney test. A 2-tailed P value <0.05 was considered significant, except for the analysis of the indirect items where a Bonferroni correction was applied for multiple comparisons. The analysis was performed using SPSS 16.0 for Windows (SPSS, Inc., an IBM Company, Chicago, Illinois).

Results

From the 393 questionnaires that were sent out, a total of 195 completed forms were returned, with 118 received after the first round, 23 forms after the second round, and another 54 after the third round. Seven forms were returned undeliverable. Hence, the overall response rate was 50.5%.

The physicians who replied were mostly male (63%) and resident (57%) with 64% internal medicine specialists, 32% surgeons, and 4% emergency physicians. Median age was 31 (SD 8.9) years (range, 25–64 years). Gender, discipline, and position did not differ significantly between respondents and nonrespondents or between early and late respondents. Means, standard deviations, internal consistency reliabilities, and relations of the study variables are presented in Tables 1 and 2. The initial PBC construct had insufficient internal consistency (Cronbach’s α = 0.552). A satisfactory value of 0.853 was obtained after excluding question 25 (“Whether I use the local guidelines for my patients during next year is entirely up to me”) and question 26 (“The decision to use the local guidelines for my patients during next year is beyond my control”) from the construct. There were significant correlations between gender and position (Cramer’s V = 0.366; P < 0.001), with more women present among the residents, and between position and discipline (V = 0.186; P = 0.037), with more residents in the internal medicine subgroup. No association was found between gender and discipline (V = 0.162; P = 0.085).

Table 1

Main Statistics of Studied Variables

	No.^a	Mean^b (SD)	Median^b	Skewness
Intention (3 items)	195	6.44 (0.82)	7.00	−2.65
Attitude (6 items)	195	5.86 (1.05)	6.00	−1.54
Subjective norms (4 items)	195	4.78 (1.15)	4.75	−0.47
Perceived behavioral control (3 items)^c	193	6.06 (0.75)	6.00	−0.47
Habit strength (12 items)	195	4.49 (1.07)	4.58	−0.21
Frequency^d	190	2.65 (1.18)	3.00	−0.12
Age	187	35.22 (8.90)	31.00	1.17
Discipline^e	191	Surgery	Internal medicine	Emergency medicine
		61 (32)	122 (64)	8 (4)
Position^e	191	Staff	Resident
		82 (43)	109 (57)
Gender^e	188	Male	Female
		118 (63)	70 (37)

Number of valid responses.

Minimum mean score = 1; maximum mean score = 7.

Excluded items = question 25 (“Whether I use the local guidelines for my patients during the next year is entirely up to me”) and question 26 (“The decision to use the local guidelines for my patients during the next year is beyond my control”).

Average prescribing frequency of antibiotics per week: 1 = never; 2 = 1–2 times; 3 = 3–4 times; 4 = daily.

Given as No. (%).

Table 2

Correlations and Internal Consistencies

	Intention	Attitude	Attitude–Indirect	SN	SN–Indirect	PBC	PBC–Indirect	Habit Strength	Age	Frequency	Gender	Position
Intention	(.945)^a
Attitude	.129	(.817)^a
Attitude indirect	.201*	.287**	(.605)^a
SN	.140	.183*	.165*	(.730)^a
SN indirect	.251**	.213**	.302**	.654**	(.715)^a
PBC (3 items^b)	.260**	.211**	.448**	.361**	.419**	(.853)^a
PBC indirect	.196**	.268**	.455**	.349**	.354**	.437**	(.735)^a
Habit strength	.267**	.149*	.287**	.413**	.473**	.357**	.470**	(.914)^a
Age	.010	−.090	.069	.070	.010	.142	.193**	.105	—
Frequency^c	−.084	−.160*	.109	−.047	−.174*	−.053	.081	.030	.029	—
Gender^d,e	.122	.170	.082	−.008	.132	.007	−.079	0.097	−.476**	−.125	—
Position^d,f	.091	.078	−.062	−.040	.036	−.118	−.109	−.050	−.999**	.006	—	—

Correlations with gender and position are point-biserial; other correlations are given as Pearson’s r. SN, subjective norm; PBC, perceived behavioral control.

Internal consistency (Cronbach’s alpha) for composite variables (attitude, subjective norm, perceived behavioral control, habit); N = 184.

Average prescribing frequency of antibiotics per week: 1 = never; 2 = 1–2 times; 3 = 3–4 times; 4 = daily.

Point-biserial correlation.

1 = male; 2 = female.

1 = staff member; 2 = resident.

P < 0.05. **P < 0.01.

Regression Analysis

The sociodemographic variables, entered in the first step of the multiple regression analysis, did not explain a major portion of the variance in intention, although position and age had a significant influence that disappeared once other predictors were included (Table 3). As a set, the TPB predictors, entered in step 2, accounted for a significant amount of additional variance in the explanation of intention. Only PBC was significant, however (P = 0.005). After including HAB, the final model accounted for 13.4% of total intention variability. In this model, HAB had the largest significant influence (relative weight = .391), closely followed by PBC (relative weight = .354). ATT and SN were not significant throughout the analysis.

Table 3

Hierarchical Regression Analysis: Standardized Regression Coefficients and R²

Variables	Model 1 (R² = .041)	Model 2 (R² = .104)	Final Model (R² = .134)	Relative Weight
Position^a	.257*	.233	.223	.035
Age	.244*	.189	.163	.037
Gender^b	.079	.060	.039	.068
Frequency^c	−.086	−.063	−.077	.032
Attitude	—	.058	.048	.043
Subjective norm	—	.037	−.027	.040
PBC^d	—	.221**	.178*	.354
Habit	—	—	.197*	.391

N = 184. Relative weight for each variable expressed as a fraction of R²·²⁴ PBC, perceived behavioral control.

1 = staff member; 2 = resident.

1 = male; 2 = female.

Average prescribing frequency of antibiotics per week: 1 = never; 2 = 1–2 times; 3 = 3–4 times; 4 = daily.

Three-item construct used for analysis; excluded items = question 25 (“Whether I use the local guidelines for my patients during the next year is entirely up to me”) and question 26 (“The decision to use the local guidelines for my patients during the next year is beyond my control”).

P < 0.05. **P < 0.01.

The position parameter reached significance in the first model only (P = 0.04) but remained marginally nonsignificant in succeeding steps (second model, P = 0.051; final model, P = 0.057). This suggested a possible moderating effect of position. Therefore, a subgroup analysis was undertaken for residents and staff members (Table 4). In the resident subgroup, intention was significantly influenced only by PBC (relative weight = .426; model R² = .141), whereas in the staff subgroup, HAB was the single significant predictor (relative weight = .470; model R² = .152). A similar subgroup analysis between internal medicine and surgery showed no differences.

Table 4

Regression Analysis Models with Position as Moderator: Standardized Regression Coefficients and R²

Position	Variables	Model 1 (R² = .053)	Model 2 (R² = .092)	Final Model (R² = .152)	Relative Weight
Staff member (n = 79)	Age	.218	.155	.151	.170
	Gender^a	.113	.099	.125	.068
	Frequency^b	−.064	−.025	−.049	.021
	Attitude	—	.015	.020	.017
	Subjective norm	—	−.024	−.111	.028
	PBC^c	—	.212	.107	.226
	Habit	—	—	.295*	.470
	Independent Variables	Model 1 (R² = .020)	Model 2 (R² = .130)	Final Model (R² = .141)	Relative Weight
Resident (n = 105)	Age	−.016	.027	.016	.001
	Gender	.042	.024	−.002	.014
	Frequency	−.129	−.107	−.116	.068
	Attitude	—	.114	.101	.157
	Subjective norm	—	.096	.053	.108
	PBC^c	—	.244*	.228*	.426
	Habit	—	—	.123	.226

Standardization of the coefficients has been done within each separate group. Relative weight for each variable expressed as a fraction of R²·²⁴ PBC, perceived behavioral control.

1 = male; 2 = female.

Average prescribing frequency of antibiotics per week: 1 = never; 2 = 1–2 times; 3 = 3–4 times; 4 = daily.

P < 0.05.

Indirect Measures

The results of the indirect items are presented in Table 5 (see Web appendix). Scores for attitude-related items were generally positive toward the use of guidelines in daily practice. Regarding the question if guidelines could lead to more work, men were slightly more concerned than women (median score 3.0 v. 2.0; P = 0.002). For subjective norm topics, physicians gave rather moderate ratings, indicating limited positive social pressure. No major differences were found when comparing position, gender, and disciplines. For perceived behavioral control items, questions concerning the amount of attention toward guidelines during training and on the wards received low ratings, with surgeons giving lower scores than internal medicine (median scores on a 7-point scale expressing agreement: training, 2.0 v. 3.0; wards, 3.0 v. 4.0; both P = 0.002). Feedback and the amount of received support scored equally low.

On the question who should give additional antibiotic guidance, the responding physicians gave the infectious disease specialist the highest approval rating (median agreement score: 5.1 on the 1–7 scale), with lower ratings for the microbiologist (median score 4.9) and the clinical pharmacist (median score 4.5). These differences were highly significant (Friedman’s χ², P < 0.001) even between infectious diseases and microbiology (Wilcoxon signed ranks; P = 0.009). No differences were present among gender, discipline, or position.

Discussion

In the present study, we find that physicians’ intention to use antibiotic guidelines is influenced by habits and by their perceived control over the use of the guidelines and that the determinants of physician intention to follow guidelines are different for residents and faculty. It is important to note that only 13% of the variability was accounted for, so the findings of this survey must be taken cautiously. Also, neither attitude nor subjective norms proved to be determinants influencing final intention. This is in contrast with older literature where personal opinions on guidelines are quoted as important barriers.⁶

One explanation for the lack of influence of attitude and subjective norms may be the specific site of this study, a university hospital with its training function. As has been described previously, physicians in a teaching hospital have more positive opinions on guidelines than those from nonteaching hospitals,²⁵ and hospital physicians have higher attitude scores than primary care physicians.²⁶ The physicians in our study had a generally positive attitude toward antibiotic guidelines, as well as a positive intention to follow the guidelines, each with a relatively small variance. As such, there is limited opportunity for attitude to influence intention.²⁷ Another possibility is the effect of recent national campaigns on antibiotic use. These have already led to a decrease in outpatient and hospital antibiotic consumption in Belgium.²⁸ Potentially, these campaigns may have increased awareness of existing guidelines or could have induced a more positive view of directives concerning antimicrobial use. Finally, socially desirable answers may have decreased variability.

A similar study on antibiotic prescribing for a sore throat already showed the importance of PBC and habits.¹⁴ There, the authors demonstrated that past behavior and control beliefs are highly influential parameters in the intention to use antibiotic guidelines. However, they questioned the effect of past behavior because of the unreliability of self-report in their measure. Measuring habit strength is less vulnerable to this effect, and our study confirms that habits indeed play a significant role in physicians’ intention to use antibiotic guidelines. This influence of habits has important consequences for the use and spread of new evidence and guidelines. Changing practice will involve both removing old habits and the formation of new habits more consistent with guidelines.

The moderator effect of position is similar to previous observations where younger and less experienced physicians are influenced by guidelines to a greater degree than their older and more experienced counterparts.^29,30 The consequence of this finding is that to improve guideline use among residents, the main focus of an intervention should be on improving the direct availability of the guidelines combined with enhancing self-efficacy. This can be done by an easy-to-use guideline format, more applied education on and familiarization with the guidelines, and feedback and guidance on how the local guidelines are actually devised and used. Creating the habit of using guidelines should be the aim for this group because future practice will be influenced by it. If possible, these interventions should be planned when residents start at a new location because a change in environment provides an opportunity to create a new habit while it also disrupts old habits.³¹ For staff members, the focus should be on changing old habits. Interventions such as improved guidance or different education and guideline formats (so-called downstream interventions) may be less successful for staff members.³¹ To promote new habits, interventions that target the environment are needed (upstream interventions). Possibilities include the use of automated decision support systems such as TREAT,³² providing direct suggestions and recommendations to the physician that may initiate a new habit.³³ Enhanced support from other disciplines can be an alternative. Policy changes and economic incentives may also induce stable changes.³¹ The most influential guidance may come from the infectious disease specialists as they have the highest “street” credibility, as previously observed.⁸ If microbiology or clinical pharmacy is to be involved in these interventions, their role will have to be clearly defined and supported.

Attitudes and subjective norms are found to be of lower influence in our situation. Nevertheless, they should not be entirely neglected: Every specific intervention should mention the importance of guidelines and support from peers, and supervisors will remain a determinant for success.

Limitations

The most important limitations in our study are a low explained variability, a ceiling effect on intention and attitude, and the fact that we cannot rely on actual measures of past and current behavior. First, the variables measured explained only a small portion of the variability of the intention to follow guidelines (13%), even lower for the separate variables (e.g., habit strength and PBC each explain only between 5% and 7% of total intention variance). Questionnaires are susceptible to socially desirable answering, especially on sensitive topics such as attitude and intention regarding antibiotics, and with low-power respondents such as trainees. Furthermore, given the response rate of 50%, our sample may be biased toward physicians with a more favorable attitude toward the hospital, further reducing explained variability as well as the generalizability of our findings. This ceiling effect may account for the low proportion of intention variance that was explained. However, this may also reflect our current understanding of the determinants of adherence to antibiotics guidelines.

Our study did not measure actual past or current behavior because determining antibiotic guideline use for each participating physician would be not feasible in our hospital. Therefore, we choose to rely only on intention, which is normally a good predictor of the actual behavior and measured by a validated method.¹¹ Similarly, reported habit strength was used to estimate past behavior. However, measuring the actual behavior will be needed to confirm our observations.

The construct for measuring PBC showed some problems with internal consistency, leading to a need to omit some items. Close attention should be paid to this construct in the future. Also, the specifically adapted questionnaire and the fact that our hospital is a major university hospital may limit its use in other settings. More studies should be conducted to fully define the role of hierarchical position, habits, and other behavioral determinants in the use of antimicrobial guidelines.

Conclusions

To improve physicians’ antimicrobial practice, knowledge of the different barriers to and facilitators of guideline adherence, together with their relative importance, is needed to select the most appropriate interventions. Our questionnaire, based on the theory of planned behavior and adapted to the local situation, was able to clarify differences between residents and staff members in how physicians’ intention on guideline use is formed. Although staff members experience influence of previous routine and habits, residents are guided by external influences and how much control they experience. Most important, these divergent origins suggest that different approaches to improving antimicrobial use may be necessary. For staff members, methods that focus on breaking habits and giving direct proposals such as automated decision support systems are possible interventions. Residents, on the other hand, may be best targeted using convenient guideline formats, familiarization and guidance on how the local guidelines are actually devised and used, and feedback.

Footnotes

Acknowledgements

The authors thank all physicians who participated in the pilot and the final survey. They also acknowledge Prof. Dr. P. Broos, Department of Surgery, Prof. Dr. J. B. Gillet, Emergency Department, Prof. Dr. W. Robberecht, Department of Neurology, and Prof. Dr. C. Van Geet, Department of Paediatrics, from the University Hospitals Leuven, for their active support. Finally, the authors thank Prof. Dr. F. Riou, Université de Rennes, France, and Prof. Dr. C. Limbert, University of Wales, UK, for providing the original questionnaires.

This study was supported by an unconditional grant from the Flemish Society of Hospital Pharmacists (VZA), Belgium.

Supplementary material for this article is available on the Medical Decision Making Web site at .

References

Dellit

Owens

McGowan

Jr . Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America guidelines for developing an institutional program to enhance antimicrobial stewardship. Clin Infect Dis. 2007;44(2):159–77.

Nathwani

. Antimicrobial prescribing policy and practice in Scotland: recommendations for good antimicrobial practice in acute hospitals. J Antimicrob Chemother. 2006;57(6):1189–96.

Davey

Brown

Fenelon

. Interventions to improve antibiotic prescribing practices for hospital inpatients. Cochrane Database Syst Rev. 2005;(4):CD003543.

McCabe

Kirchner

Zhang

Daley

Fisman

. Guideline-concordant therapy and reduced mortality and length of stay in adults with community-acquired pneumonia: playing by the rules. Arch Intern Med. 2009;169(16):1525–31.

Arnold

LaJoie

Brock

. Improving outcomes in elderly patients with community-acquired pneumonia by adhering to national guidelines: Community-Acquired Pneumonia Organization International cohort study results. Arch Intern Med. 2009;169(16):1515–24.

Cabana

Rand

Powe

. Why don’t physicians follow clinical practice guidelines? A framework for improvement. JAMA. 1999;282(15):1458–65.

Schouten

Hulscher

Trap-Liefers

. Tailored interventions to improve antibiotic use for lower respiratory tract infections in hospitals: a cluster-randomized, controlled trial. Clin Infect Dis. 2007;44(7):931–41.

Cortoos

De Witte

Peetermans

Simoens

Laekeman

. Opposing expectations and suboptimal use of a local antibiotic hospital guideline: a qualitative study. J Antimicrob Chemother. 2008;62(1):189–95.

Mol

Rutten

Gans

Degener

Haaijer-Ruskamp

. Adherence barriers to antimicrobial treatment guidelines in teaching hospital, the Netherlands. Emerg Infect Dis. 2004;10(3):522–5.

10.

Michie

Johnston

Francis

Hardeman

Eccles

. From theory to intervention: mapping theoretically derived behavioural determinants to behaviour change techniques. Appl Psychol. 2008;57(4):660–80.

11.

Ajzen

. Constructing a TpB Questionnaire: Conceptual and Methodological Considerations; 2002 (revised 2006) [cited December 22, 2010]. p 1–13. Available from: URL: http://people.umass.edu/aizen/pdf/tpb.measurement.pdf

12.

Perkins

Jensen

Jaccard

. Applying theory-driven approaches to understanding and modifying clinicians’ behavior: what do we know? Psychiatr Serv. 2007;58(3):342–8.

13.

Limbert

Lamb

. Doctors’ use of clinical guidelines: two applications of the theory of planned behaviour. Psychol Health Med. 2002;7(3):301–10.

14.

Walker

Grimshaw

Armstrong

. Salient beliefs and intentions to prescribe antibiotics for patients with a sore throat. Br J Health Psychol. 2001;6(pt 4):347–60.

15.

Lambert

Salmon

Stubbings

Gilomen-Study

Valuck

Kezlarian

. Factors associated with antibiotic prescribing in a managed care setting: an exploratory investigation. Soc Sci Med. 1997;45(12):1767–79.

16.

Verplanken

Orbell

. Reflections on past behavior: a self-report index of habit strength. J Appl Soc Psychol. 2003;33(6):1313–30.

17.

Francis

Eccles

Johnston

. Constructing Questionnaires Based on the Theory of Planned Behaviour: A Manual for Health Services Researchers; 2004 [cited December 22, 2010]. p 1–42. Available from: URL: http://www.rebeqi.org/ViewFile.aspx?itemID=212

18.

Verplanken

. Beyond frequency: habit as mental construct. Br J Soc Psychol. 2006;45:639–56.

19.

Francis

Johnston

Eccles

Grimshaw

Kaner

EFS

. Measurement Issues in the Theory of Planned Behaviour: A Supplement to the Manual for Constructing Questionnaires Based on the Theory of Planned Behaviour; 2004 [cited December 22, 2010]. p 43-67. Available from: URL: http://www.rebeqi.org/ViewFile.aspx?itemID=219

20.

AGREE Collaboration. Development and validation of an international appraisal instrument for assessing the quality of clinical practice guidelines: the AGREE project. Qual Saf Health Care. 2003;12(1):18–23.

21.

Funk

Champagne

Wiese

Tornquist

. BARRIERS: the barriers to research utilization scale. Appl Nurs Res. 1991;4(1):39–45.

22.

Riou

Veillard

Jarno

Chaperon

Coutte

. Clinical guidelines: what are the viewpoints of physicians and housestaff working in hospital medical departments? Rev Epidemiol Sante Publique. 2005;53(Spec No 1):1S57–66.

23.

Netemeyer

Bearden

Sharma

. Scaling Procedures: Issues and Applications. Thousand Oaks, CA: Sage; 2003.

24.

Johnson

. A heuristic method for estimating the relative weight of predictor variables in multiple regression. Multivariate Behav Res. 2000;35(1):1–19.

25.

Greving

Denig

de Zeeuw

Haaijer-Ruskamp

. Physicians’ attitudes towards treatment guidelines: differences between teaching and nonteaching hospitals. Eur J Clin Pharmacol. 2006;62(2):129–33.

26.

Formoso

Liberati

Magrini

. Practice guidelines: useful and “participative” method? Survey of Italian physicians by professional setting. Arch Intern Med. 2001;161(16):2037–42.

27.

Ajzen

Fishbein

. The influence of attitudes on behavior. In: Albarracin

Johnson

Zanna

MPE

, eds. The Handbook of Attitudes. Mahwah, NJ: Lawrence Erlbaum; 2005. p 173–221.

28.

Goossens

Coenen

Costers

. Achievements of the Belgian Antibiotic Policy Coordination Committee (BAPCOC). Euro Surveill. 2008;13(46):pii 19036.

29.

Sammer

Lykens

Singh

. Physician characteristics and the reported effect of evidence-based practice guidelines. Health Serv Res. 2008;43(2):569–81.

30.

Halm

Atlas

Borowsky

. Understanding physician adherence with a pneumonia practice guideline: effects of patient, system, and physician factors. Arch Intern Med. 2000;160(1):98–104.

31.

Verplanken

Wood

. Interventions to break and create consumer habits. J Publ Pol Market. 2006;25(1):90–103.

32.

Leibovici

Paul

Nielsen

Tacconellic

Andreassen

. The TREAT project: decision support and prediction using causal probabilistic networks. Int J Antimicrob Agents. 2007;30:S93–S102.

33.

Wood

Neal

. A new look at habits and the habit-goal interface. Psychol Rev. 2007;114:843–63.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB