A prospective study of patients’ pain intensity after cardiac surgery and a qualitative review: effects of examiners’ gender on patient reporting

2.1 Ethics and registration

This study was performed between 2012 and 2015 at the University Hospital Bergmannsheil in Bochum, Germany. The local Ethics Committee of the Ruhr University approved the study protocol (No. 4484-12), which was in accordance with the Declaration of Helsinki. Written informed consent was obtained from all patients before enrolment. The study was registered on German Clinical Trials Register (DRKS00011597).

2.2 Characteristics of participants

Female (f) and male (m) patients (40–90 years) scheduled for elective heart surgery via median sternotomy were included. Exclusion criteria were any psychiatric disorders [e.g. known depression or BDI-II score ≥29 (see below)], patients taking any pre-existent opioids (including e.g. codeine, tramadol, piritramide, tapentadol, oxycodone) indicating chronic pain due to a probable higher analgesic need postoperatively, missing language skills and dementia, as well as declination of participation. Postoperative exclusion criteria were ongoing analgosedation needed for mechanical ventilation, delirium or revocation of previous given consent. Participants were told that the aim of the study was to survey postoperative pain intensity in order to improve analgesic therapy. Patients and assessors (students) were blinded to the study hypothesis.

2.3 Characteristics of examiners and group affiliation

The assessors were four students of both genders: two females (F) and two males (M) (aged 24.3 years±1.7). All four shared an identical cultural background and ethnicity with hair ranging from blonde to dark blonde and wore doctors’ scrubs.

At the first postoperative assessment, patients were allocated consecutively to one of the four assessors (here called assessor or examiner). This group allocation stayed consistent until the end of the study. One at a time, each assessor collected the data of patients with the same and the opposite gender. Therefore, the following study groups were defined:

1. Group “Ff”: female examiners – female patients;

2. Group “Fm”: female examiners – male patients;

3. Group “Mf”: male examiners – female patients;

4. Group “Mm”: male examiners – male patients.

Each patient was only consulted by one assessor (male or female). This was done intentionally not to influence the reported pain intensity by being asked twice in quick succession or having any circumstances between the two assessments which could have caused different pain ratings (e.g. physiotherapy, redressing wounds or even coughing).

2.4 Study design

Included patients were visited on the evening before the elective surgery and were asked to answer standardized questions about their demographic factors and medical backgrounds and to complete two validated questionnaires to check for psychological variables like anxiety and depression: the “State and trait anxiety inventory” (STAI) and “Becks Depression Inventory” (BDI-II); assistance was offered at all stages by the examiners.

STAI lists 20 state-anxiety (STAI-S; feeling at the particular moment) and 20 trait-anxiety items (STAI-T; general feeling), each must be rated on a four-point intensity scale (from 1: “not at all”, to 4: “very much so”). The range of scores for each subtest is 20–80; higher scores indicate a tendency to suffer from anxiety when in an anxiety inducing environment (STAI-T) or more anxiety symptoms (STAI-S): scores on STAI-S and STAI-T ≥30 suggest moderate anxiety and scores ≥45 suggest severe anxiety [18].

BDI-II is a 21-question self-rating scale for measuring the severity of clinical depression experienced during the previous 2 weeks [19]. Each answer is scored on a scale value of 0–3. Higher total scores indicate more severe depressive symptoms. Patients with a score ≥29 preoperatively, indicating a severe depression, were excluded.

Both, anxiety and depression, were investigated as variables in the tested hypothesis for examiners’ gender influences.

During preoperative assessment, patients were coached to report their current postoperative pain intensity on a Numerical Rating Scale (NRS) (0: “no pain”, 10: “worst imaginable pain”) and to rate quality of pain using a German validated list of verbal descriptors (“Schmerzempfindungsskala”; SES) [20]. The SES is a standard instrument encompassing 10 descriptors on sensory and 14 descriptors on affective subscales. Patients rated every descriptor on a four-level categorical scale (1: “not appropriate”, 4: “fully appropriate”), and both, a sensory and an affective sum score were built. A numerically high value means a high level of pain on the respective scale (the higher the sum value, the more intense the perception or the higher the emotional burden due to the acute pain).

Additionally, a score for delirium was assessed on the five-item Nursing Delirium Screening Scale (NUDESC) [21]. Symptoms are rated from 0 to 2 based on the presence and intensity of each symptom (disorientation, inappropriate behavior, inappropriate communication, illusions/hallucinations, psychomotor retardation; maximal score of 10). Sums equal to, or ≥2, show an acute delirium.

Using a standardized study protocol, patients were asked for postoperative nausea and vomiting, their current pain intensity on NRS (“how is your current pain intensity while resting on ICU”), pain quality by SES and delirium by NUDESC, twice on the first and twice on the second postoperative day (POD, approx. 09:00: POD_1.1/POD_2.1 and approximately 17:00: POD_1.2/POD_2.2). In the standardized clinical routine, the current pain intensity while resting provides the fundament for further pain management if necessary. For all patients, according to the German Clinical Practice Guidelines, a standardized regime for analgesia (implemented for years) by means of metamizole (Novalgin^® 1 g per os every 6 h) and oxycodone/naloxone (Targin^® 20/10 mg every 12 h) on a fixed time pattern was used [22], [23]. In cases of severe acute pain states rescue medication could be requested on ICU via intravenous access (e.g. 3 mg piritramide intravenously). Besides, dipyrone is the most preferred and used nonopioid analgesic in Germany. Compared with NSAID, dipyrone does not heighten the risk for gastrointestinal bleeding, thrombosis or severe renal failure. Under regular blood analysis or short duration of use, the risk of agranulocytosis remains low. However, the debate has been controversial for over 30 years [22], [23].

2.5 Statistical analysis

Using G*power version 3.1.9.3 for Macintosh, a priori power analysis revealed a total size of at least 126 patients (p<0.05, NRS mean 3, SD±2) for the outcome pain rating on NRS (i.e. at least 63 patients per investigator gender). Data were analyzed with Statistical Package for Social Sciences Software (SPSS Version 22, IBM, Chicago, IL, USA) and are expressed as mean±standard deviation (SD) for quantitative variables, and as number (percent) for qualitative variables. As no normal distribution was given for pain parameters after performing Shapiro-Wilk-test, non-parametric tests were used. To compare the effects of assessor gender or patient gender on pain reporting, Mann-Whitney-U-tests were performed. To compare the four study groups, a Kruskal-Wallis-test with Bonferroni correction was used with a post hoc Mann-Whitney-U-test. Correlations between NRS pain intensity and scores of STAI-S or BDI-II were performed by two-tailed bivariate Spearman correlations. A priori planned subgroup analyses were performed for severe anxiety levels ≥45 on STAIT-S, moderate depression (BDI-II score of 20–28), any state of depression (BDI-II score of 9–28), different levels of pain intensity (low: NRS ≤3, moderate: NRS 4–6, severe: NRS ≥7), individual assessors and groups of assessment time points (POD_1 and _2). A significance level of p<0.05 was employed.

2.6 Qualitative review

The databases Medline (via pubmed) and Central were searched for appropriate articles (i.e. observational, retrospective or randomized controlled trials) in English or German until May 2018. A qualitative review of all publications was performed focusing on experimenter gender bias.

3.1 Qualitative review

Results are depicted in Table 1. Whereas four experimental studies found increased pain ratings (or a reduced pain tolerance) in front of a female experimenter [10], [13], [15], [16], four other experimental studies reported the opposite [6], [7], [8], [12]. There are only three experimental studies which were designed specifically to investigate gender reporting bias [6], [7], [8]; all three resulting in the general conclusion that pain intensity might be lower when reported to a female examiner. Therefore, the following study hypothesis was formulated a priori.

3.2 Study hypothesis

This prospective clinical study was designed to investigate whether patients after surgery report lower pain intensities when assessed by a female compared to a male assessor.

3.3 Patient characteristics

A total of 446 patients were assessed for eligibility; of these, 161 were excluded due to declined participation (39.1%), existing language barrier (20.5%), BDI-II score for severe depression (1.9%), or complaints of chronic pain (21.7%). One hundred and twenty were excluded secondarily (Fig. 1). Finally, data from 165 patients [66.39 years±9.63; male: 118 (m), female: 47 (f)] were analyzed. Of these, 53.9% (F, n=89) patients were interviewed by a female and 46.1% (M, n=76) by a male assessor. All patients underwent a sternotomy, due, e.g. to coronary artery bypass grafts (84.2%), aortic (14.5%) or mitral valve replacements (10.3%, Table 2). STAIT-S and STAIT-T did not differ between male and female patients (State z=–1.198, p=0.231; Trait z=–1.162, p=0.245); however, BDI-II indicated a more frequently minimal depressive mood in females than in males but without pathological significance (z=–2.802, p=0.005). Only 144 patients (87.3%) finished the assessment on POD 2.2 (Fig. 1).

Fig. 1:

Flow diagram and study design. After inclusion criteria were met, all patients were given an initial assessment on the day before surgery. On the first two postoperative days, patients were asked to rate their pain intensity in the morning and in the evening. Exclusion based on acute delirium: Scores on “Nursing Delirium Screening Scale” (NUDESC) equal to, or greater than 2. Other reasons for secondary exclusion n=24: surgery was postponed (n=10), analgosedation needed (n=13), reoperation needed (n=1). POD=postoperative day (1.1=morning of first postoperative day, 1.2=evening of first postoperative day, 2.1=morning of second postoperative day, 2.2=evening of second postoperative day). F=female examiner; f=female patient; M=male examiner; m=male patient.

3.4 Pain reporting

Summarizing all patients, pain intensity on POD_1 was rated 4.0±2.4 on NRS and decreased on POD_2 to 3.0±2.1 [H(3)=37.941, p=0.000]. Pain scores of male and female patients (if male and female assessor was neglected) showed no differences at any time after surgery (z=–0.164, p=0.869; Fig. 2). This remains consistent in each subgroup analyses for low, moderate and severe pain. The same was true for SES descriptors of pain quality between female and male patients (Table 3 and Table Supplement).

Fig. 2:

Gender-specific reporting of postoperative pain intensity. Comparing pain scores of female and male patients within 2 days after heart surgery on a Numerical Rating Scale (NRS: 0=“no pain” to 10=“worst imaginable pain”). Data are presented as mean±SD. POD=postoperative day [1.1=morning of first postoperative day (n=165), 1.2=evening of first postoperative day (n=157), 2.1=morning of second postoperative day (n=145), 2.2=evening of second postoperative day (n=144)]. Kruskal-Wallis-test (3)=37.942, p=0.000.

Female patients feeling anxious (STAI-S≥45) showed higher NRS scores (n=49; NRS mean 4.0±2.1,) than anxious males (n=44; 3.1±1.8; z=–2.183, p=0.029). In all patients, independent from gender, there was no correlation between pain intensity and preoperative anxiety (r_s=0.095, p=0.364). A positive correlation was seen between all BDI-II scores and pain intensity, independent from gender (r_s=0.156, p=0.045).

3.5 Effects of examiner gender on pain reporting

Regardless of the patients’ gender, more intense pain was reported in the presence of a female examiner (NRS 3.4±2.5) when compared to the pain reported to a male examiner on POD_2 (NRS 2.4±2.2; z=–3.673, p=0.000, Table 3). This difference remains for moderate (NRS difference 0.61; z=–2.35, p=0.019), but not for low or severe pain levels. No differences due to assessor gender were regarded on POD_1 (z=–0.02, p=0.984, Table 3). In front of a female assessor, the quality of pain was described with more intense affective (19.8% vs. 17.6%, p=0.008) and sensory components (16.6% vs. 11.9%, p=0.004) than in front of a male assessor on POD_ 2.1 (Table Supplementary).

Considering the four different study groups regarding gender of assessor and gender of patient, a significant difference was seen on POD_2 for pain intensity [H(3)=14.139, p=0.003] and SES (Table 3). In particular, male patients reported higher pain scores to a female assessor compared to a male assessor (NRS difference 1.2; z=–3.616, p=0.000; Fig. 3). This difference remains consistent for moderate (NRS difference 1.1; z=–3.06, p=0.002), but not for low or severe pain levels. In addition, female patients reported descriptively higher pain scores to a female examiner again on POD_2.1 but without reaching significance (NRS difference 1.1; z=–1.426, p=0.154).

Fig. 3:

Reporting of pain intensity on the second postoperative day depending on gender of examiners and patients. Scatter plot of NRS scores (NRS: 0=“no pain” to 10=“worst imaginable pain”) divided into gender of examiner and patients. Data are presented as mean values. Overall mean is displayed as dashed line. n=145. Comparison of the four groups revealed in Kruskal-Wallis-test [(3)=14.139, p=0.003] a main significant difference. Mann-Whitney-U-test revealed differences between Fm vs. Mf (NRS difference 0.88, z=–2.088, *p=0.037) and Fm vs. Mm (NRS difference 1.22, z=–3.616, **p=0.000). F=female examiner; f=female patient; M=male examiner; m=male patient.

Subgroup analyses investigating anxiety revealed no significant effects (Fig. 4). There were no significant effects on any measures due to individual assessors.

Fig. 4:

Heat map of pain intensity depending on gender of examiners and patients. Displayed is the percentage of patients with low or moderate (STAI-S≤44) vs. severe anxiety (STAI-S≥45) on the first and second postoperative day (POD). Each pain intensity on Numerical Rating Scale (NRS: 0=“no pain” to 10=“worst imaginable pain”) is plotted in columns. Color codes reflect low (white) and high (red) percentage (sum of frequencies on POD_1 and POD_2) of patients. n=165.

4.1 Strength and limitations

Our study is limited by only including post-sternotomy patients with a mean age of 66 years. In addition, patients scheduled for heart surgery are frequently multimorbid and are at a higher risk for postoperative complications, both may explain a higher exclusion rate compared to a young and healthier population.

We investigated confounders like depression and anxiety. Both might be linked to pain intensity, but both independent from examiner gender. As the topic of gender influences on pain intensity seems to be even more complex in the clinical environment, there are probably further unrecognized confounders.

As an a priori power analysis revealed a total size of at least 126 for the outcome pain rating on NRS, we reached our suggested threshold. However, due to a generally smaller number of females undergoing heart surgery there was a smaller sample size in the female patient group.

In order to maintain the clinical routine dosages for analgesic management, a non-randomized study design was chosen and the analgesic regime was not adjusted to mg/kg or otherwise changed. Furthermore, a double-blinded, randomized cross-over design would have been preferable, but preliminary results indicated that patients would have been influenced, on the one hand, and exhausted, on the other, if asked to repeatedly rate their pain intensity.

The strength of our study was a standardized study protocol including a standardized analgesic regime investigating a homogenous sample of patients. Additionally, we chose four instead of just one male and one female examiner to exclude individual confounders such as sympathy or antipathy.

In our clinical study, gender bias was not observed until POD_2, consistently for both, pain intensity (NRS) and pain quality (SES). It is speculative, but this effect could be due to a hangover of anaesthetics on POD_1 and a routinely delayed extubation after surgery that might mask gender bias by suppressing the mechanisms that are causative for its occurrence. Another explanation might be the rather high pain ratings on POD_1 which might be less prone to gender bias. However, in general patients suffering from severe pain were rare [POD_1 n=18 (out of 165), POD_2 n=8 (out of 144)], but mild pain itself seems not be a sensitive enough factor for gender differences.

Because only post-sternotomy patients were included, we do not know if these results can be generalized to other patient groups; however, we tried to be consistent for the surgical procedure in this first prospective clinical study on the gender experimenter bias to exclude an additional confounder. Furthermore, we cannot exclude that our assessors’ inexperienced professional status might have influenced the results as seen in [16]. As the students looked very young, they may have possibly seemed inexperienced. Further studies need to address these limitations.