Parallel sensitivity analyses examined changes in the percentage of respondents willing to choose antibiotics alone as different consequences of this choice were systematically improved. IV indicates intravenous.
eAppendix. Survey Scenario
eTable. Summary Table Used to Ground Sensitivity Analysis
Customize your JAMA Network experience by selecting one or more topics from the list below.
Hanson AL, Crosby RD, Basson MD. Patient Preferences for Surgery or Antibiotics for the Treatment of Acute Appendicitis. JAMA Surg. 2018;153(5):471–478. doi:10.1001/jamasurg.2017.5310
Which treatment will patients with uncomplicated acute appendicitis choose when presented with both surgical and nonsurgical options, and what might make nonsurgical antibiotic therapy a more appealing option?
In a survey of 1728 participants, most respondents chose laparoscopic appendectomy over open appendectomy and antibiotics alone as a treatment option for acute appendicitis, while a sensitivity analysis involving 220 individuals demonstrated that antibiotics alone was more appealing when short-term failure or long-term recurrence rates were decreased.
Patients should be presented with all viable treatment options, including antibiotics alone, while future research on antibiotic treatment for appendicitis should focus on improving failure rates.
Studies have compared surgical with nonsurgical therapy for acute uncomplicated appendicitis, but none of these studies have a patient-centered perspective.
To evaluate how patients might choose between surgical and nonsurgical therapy for acute uncomplicated appendicitis and to identify targets to make antibiotic treatment more appealing.
Design, Setting, and Participants
This study comprised an online survey and an in-person sensitivity analysis survey. For the web survey, a convenience sample of 1728 respondents were asked to imagine that they or their child had acute uncomplicated appendicitis, provided information about laparoscopic and open appendectomy and antibiotic treatment alone, and asked which treatment they might choose. The web survey was open from April 17, 2016, through June 16, 2016, and was disseminated via email link, a poster with a Quick Response code, and social media. For the sensitivity analysis, 220 respondents were given the same scenario and options. Those who chose surgery were asked whether certain factors influenced their decision; each factor was incrementally improved during questioning about whether respondents would consider switching to antibiotics. These participants were recruited at public venues from June 3, 2016, to July 31, 2016. Web survey data were analyzed from June 17, 2016, to September 21, 2017. Sensitivity analysis data were analyzed from August 1, 2016, to September 21, 2017.
Main Outcomes and Measures
Among the 1728 web survey respondents, 1225 (70.9%) were female and 500 (28.9%) were male (3 [0.2%] either did not answer or responded as “gender fluid” within the comments section of the survey), and most self-reported being between 50 and 59 years of age (391 [22.6%]) and being non-Hispanic white (1563 [90.5%]). For themselves, 1482 respondents (85.8%) chose laparoscopic appendectomy, 84 (4.9%) chose open appendectomy, and 162 (9.4%) chose antibiotics alone. For their child, 1372 respondents (79.4%) chose laparoscopic appendectomy, 106 (6.1%) open appendectomy, and 250 (14.5%) antibiotics alone. Respondents were somewhat more likely to choose antibiotics for themselves if they had education beyond college (105 [12.6%]; P < .001), identified as other than non-Hispanic white (24 [14.9%]; P < .001), or did not know anyone who had previously been hospitalized (12 [15.8%]; P = .02), but they were less likely to choose antibiotics if they were surgeons (11 [5.4%]; P = .008). Of the 220 participants interviewed for the sensitivity analysis, 120 (54.5%) were female and 100 (45.5%) were male, and most self-reported being between 18 and 24 years of age (53 [24.1%]) and being non-Hispanic white (204 [92.7%]). Their responses suggested that improvements in the short- and long-term failure rate of antibiotic treatment—rather than reductions in the duration of hospitalization or antibiotic treatment—were more likely to increase the desirability of choosing antibiotics.
Conclusions and Relevance
Most patients may choose surgical intervention over antibiotics alone in treatment of acute uncomplicated appendicitis, but a meaningful number may choose nonoperative management. Therefore, from a patient-centered perspective, this option should be discussed with patients, and future research could be directed at reducing the failure and recurrence rates of antibiotic treatment for appendicitis.
Appendicitis was traditionally managed surgically, barring severe resource limitations or comorbidity. Laparoscopy and percutaneous drainage have refined the algorithm but reinforced the perceived need for invasive intervention. However, uncomplicated appendicitis may be treated with antibiotics alone, albeit with different outcomes than surgery.1-5
Further trials are planned and reported results vary, but many patients with uncomplicated acute appendicitis prosper with antibiotics alone, although they risk longer hospitalization and outpatient treatment, substantial early and late failure, and potentially more complications if antibiotics fail. Physicians debate which option is “better,” but the preferred outcome should depend on patients’ values.6,7 Informed consent doctrine emphasizes providing information that generic reasonable patients would want, but not all “reasonable” people make the same choices8-11 when considering not only the probabilities of risks and benefits but also how patients value them.
We sought to investigate how patients with acute uncomplicated appendicitis might choose treatment. We conducted an anonymous internet-based survey of 1728 individuals, asking respondents to imagine that they had acute uncomplicated appendicitis; randomly presenting laparoscopic appendectomy, open appendectomy, and antibiotics alone as treatment options; and summarizing informed consent information. We asked survey respondents to choose a treatment, explored their rationale, and recorded demographics and their propensity for risky health-and-safety activities.12 We also asked how respondents would choose for a child for whom they were responsible. Because few respondents chose antibiotics, we conducted a sensitivity analysis. For this second survey, we interviewed 220 individuals for whom we systematically varied antibiotic protocols and success to identify targets to make antibiotics for appendicitis a more attractive option.
In 2 surveys, respondents were asked to imagine they had appendicitis; asked to choose among the treatment options of laparoscopic appendectomy, open appendectomy, and antibiotics alone; and given descriptions of the options and outcomes (eAppendix in the Supplement). Participation was voluntary and without compensation. We estimated risks and outcomes for each alternative from recent publications, seeking representative statistics. Each survey was iteratively tested and refined on medical and nonmedical personnel, students (including high school students), and lay personnel. Survey language was adjusted to respond to uncertainties or respondent questions before survey release. Both surveys were approved by the University of North Dakota (UND) institutional review board. Written or electronic patient informed consent was obtained from all participants. Web survey data were analyzed from June 17, 2016, to September 21, 2017. Sensitivity analysis data were analyzed from August 1, 2016, to September 21, 2017.
We disseminated to a convenience sample of participants an anonymous web-based survey via email links, a poster with a Quick Response code, and social media. The UND licenses the survey software used (Qualtrics; Qualtrics, LLC) and thus made no additional payment for this study. The survey was publicized via the UND and UND School of Medicine and Health Sciences newsletters, posters on the UND campus, LinkedIn, Facebook, ResearchGate, the American College of Surgeons discussion board, and the Association of VA Surgeons 2016 annual meeting. Potential respondents were encouraged to forward the link to their own networks. The survey was open from April 17, 2016, through June 16, 2016.
The relatively high proportion of surgeons (205 [11.9%]) among the respondents likely reflects the social network of our senior author (M.D.B). However, to control for potential bias, we analyzed data including and excluding surgeons or respondents with any health care employment or studies and found similar results. Lack of web access, younger than 10 years of age, and inability to understand or participate represented the exclusion criteria. An electronic consent screen required completion before the introductory scenario screen presented the options in random order and treatment could be chosen.
A 1-to-5 Likert-type scale allowed participants to independently rate the importance of each factor in their decision. They were not asked to provide a relative rank order but rather to rate the importance of each factor on its own. Respondents could, if desired, rate all 5 items similarly. Factors included quick treatment, pain, avoiding surgery, avoiding recurrence, and avoiding complications. Patients might conceivably consider many diverse factors in making this decision, which is worthy of further research in itself. To limit survey length and respondent fatigue, we chose these 5 factors on the basis of our senior author’s experience in offering these choices to patients in actual practice and on the questions that patients have asked. Respondents were then asked to imagine that their child, or a child for whom they were responsible, had acute uncomplicated appendicitis instead, to choose for this child, and to describe what influenced their decision.
We recorded respondents’ gender, education, race or ethnicity, age group, and occupation as surgeon as well as whether they had children or previous experiences with surgery, hospitalization, and illness. Participants were asked to rate their likelihood of participation in 8 health-risk behaviors using a previously validated assessment of health-risk behavior.12 These behaviors included the likelihood of engaging in illegal drug use, consuming more than 5 alcoholic drinks per day, having unprotected extramarital sex, riding a motorcycle without a helmet, not wearing a seatbelt in the front seat, getting sun exposure without sunscreen, walking home alone at night in a somewhat unsafe area, and regularly eating high-cholesterol foods.
We used χ2 analyses for categorical measures or 1-way analysis of variance for continuous measures to evaluate univariate associations between choice and demographics, personal health history, and health-risk behavior. Univariate factors associated with choice at P ≤ .10 were entered into a multivariate multinomial logistic regression predicting treatment choice. We analyzed self-choice separately from child choice.
Because few respondents in the first survey chose antibiotics, we conducted a second survey in which we interviewed 220 separate participants. We systematically improved the risks and outcomes of antibiotic treatment for this group to better understand what would influence respondents to choose antibiotics. Participants were recruited from June 3, 2016, to July 31, 2016, at public venues throughout Grand Forks, North Dakota, including the UND campus, churches, and businesses. After giving consent, participants read an introductory handout identical to that provided in the first survey (eAppendix in the Supplement) and received a summary (eTable in the Supplement) to ground the main scenario as we hypothetically varied antibiotic outcomes.
We asked respondents to imagine they had acute uncomplicated appendicitis, choose treatment based on the handout, and explain their choice. Respondents who chose surgery were asked whether certain factors influenced their decision and then a series of questions about whether they would consider switching to antibiotics; each time, we incrementally improved that specific factor. Each factor was reset to the initial survey base case before the next factor was altered. We varied these factors: length of hospitalization, duration of oral antibiotics, risk of short-term antibiotic failure, risk of long-term recurrence, and risk of developing additional complications after failure or recurrence. All participants were asked about gender, age, education, race or ethnicity, personal or family experience with appendicitis, previous surgical procedures, and hospitalization.
In total, 2153 individuals accessed the online survey, but only 1728 (80.3%) completed it (Table 1). (Virtually all others stopped at the consent or introductory scenario screens.) Among the 1728 respondents, 1225 (70.9%) were female and 500 (28.9%) were male (3 [0.2%] either did not answer or responded as “gender fluid” within the comments section of the survey), and most self-reported being between 50 and 59 years of age (391 [22.6%]) and being non-Hispanic white (1563 [90.5%]). The survey platform, which estimates the approximate location of survey respondents based on IP addresses while maintaining respondent anonymity, revealed 1123 participants (65.0% [1123 of 1728]) were from the Midwest, but responses were received from 48 states and 19 countries.
Of the 1728 respondents, 1482 (85.8%) chose laparoscopic appendectomy, 84 (4.9%) open appendectomy, and 162 (9.4%) antibiotics alone. Respondents who chose antibiotics considered avoiding surgery to be most important, and they rated quick treatment, pain, avoiding recurrence, and avoiding complications less important than did respondents who chose either open or laparoscopic appendectomy. Those who selected antibiotics rated the following from highest to lowest level of importance: avoiding surgery, avoiding complications, quick treatment, avoiding recurrence, and pain. The rating order of motivations for those who chose laparoscopic surgery matched that for those who chose open surgery; that order from most to least important is as follows: quick treatment, avoiding recurrence, avoiding complications, pain, and avoiding surgery. All results were significant at P < .002 (Table 2).
On univariate analysis, we found that respondents who chose antibiotics were more likely to have postcollege education, to not be surgeons, and to walk home alone at night in a somewhat unsafe area compared with respondents who opted for laparoscopic or open surgery. Respondents aged 60 years or older chose antibiotics less frequently than did those aged 20 to 59 years.
Multivariate analysis compared the respondents divided separately among the 3 choices, validating many univariate findings. With higher educational level, participants were more likely to choose antibiotics and less likely to choose laparoscopic or open surgery. Compared with nonsurgeons, surgeons were less likely to choose antibiotics (151 [9.9%] vs 11 [5.4%]; P < .001) and more likely to choose open surgery (67 [4.4%] vs 17 [8.3%]; P = .008). Respondents aged 20 to 59 years were more likely to choose antibiotics than those aged 60 to 79 years (144 of 1408 [10.2%] vs 14 of 288 [4.9%]; P < .001), and those who selected antibiotics were more likely to walk home alone at night than those who opted for laparoscopic surgery or open surgery (rating, 2.16 of 5 vs 1.92 of 5 vs 1.89 of 5; P = .02).
Although some trends were consistent in both analyses, some emerged in multivariate analysis. Participants were more likely to choose antibiotics instead of laparoscopic or open surgery if they self-identified as other than non-Hispanic white (24 [14.9%]; P < .001), knew no one with appendicitis (81 [10.8%]; P = .03), knew people who had undergone surgery (159 [9.5%]; P = .03), or knew no one who was hospitalized (12 [15.8%]; P = .02). All univariate and multivariate findings exhibited P ≤ .03 (Table 3). Multivariate analysis yielded otherwise similar results if health-risk behaviors were excluded from analysis.
When asked to choose for children with appendicitis, 1372 respondents (79.4%) chose laparoscopic appendectomy, 106 (6.1%) open appendectomy, and 250 (14.5%) antibiotics. Factors in decision making were rated similarly to when respondents chose for themselves (Table 2). The univariate analysis for child choice indicated that nonsurgeons and respondents who neither had appendicitis nor knew anyone who had appendicitis were more likely to choose antibiotics for their child than were surgeons or those who had or knew someone who has had appendicitis.
On multivariate analysis comparing choices for children (Table 4), respondents were more likely to choose antibiotics if they had postcollege education (137 [16.4%]; P = .005), knew nobody with appendicitis (130 [17.4%]; P = .003), or were nonsurgeons (236 [15.5%]; P = .002). Those who chose antibiotics were less likely to consume 5 or more servings of alcohol per day than those who chose laparoscopic surgery or open surgery (rating, 1.64 of 5 vs 1.85 of 5 vs 1.87 of 5; P = .025). Respondents aged 30 to 59 years more frequently chose antibiotics for their child compared with those aged 60 to 79 years (174 of 1114 [15.6%] vs 36 of 288 [12.5%]; P < .001). However, less than 19% of the respondents in any subgroup chose antibiotics for their child. All of these results were significant at P ≤ .025. Multivariate analysis yielded otherwise similar results when health-risk behaviors were excluded from analysis. We also repeated multivariate analyses for child choice using self-choice as an additional covariate. Age and education were no longer significant independent predictors, but our conclusions about other predictors were unchanged.
Interviewees’ initial decision making resembled that of the web-based survey respondents: 191 (86.8%) of 220 participants chose laparoscopic surgery, 12 (5.5%) chose open surgery, and 17 (7.7%) chose antibiotics. In general, interviewees demographically resembled web survey respondents (Table 1) except that interviewees were closer to gender balance. Of the 220 participants interviewed for the sensitivity analysis, 120 (54.5%) were female and 100 (45.5%) were male, and most self-reported being between 18 and 24 years of age (53 [24.1%]) and being non-Hispanic white (204 [92.7%]). More young adults participated in interviews, so fewer had achieved postcollege education than the respondents in the online survey. Stating their primary reasons for choosing treatment, 102 respondents (46.4%) alluded to hospitalization length, 77 (35.0%) listed immediate treatment, 77 (35.0%) identified re-infection, 75 (34.1%) mentioned invasiveness, 65 (29.6%) alluded to short- or long-term complications, 60 (27.3%) listed cosmesis, 25 (11.4%) alluded to previous surgical experience (good or bad), and only 12 (5.5%) mentioned pain.
Choices changed as consequences of antibiotics were varied (Figure). Short-term failure and long-term recurrence rates had the greatest effect on choice (Figure, D and E). Interestingly, 132 of the 203 participants (65.0%) who chose surgery said hospitalization length influenced their decision, but 70 (53.0%) of that group did not switch even if treatment was entirely outpatient. Similarly, 103 participants (50.7%) considered complication risk important, yet 69 (67.0%) of them chose surgery even if the complication rate matched that of surgery. In contrast, of the 177 participants (87.2%) who considered short-term failure important and 156 (76.8%) who considered long-term recurrence important, only 51 (28.8%) and 52 (33.3%), respectively, continued to choose surgery with improvement of antibiotic treatment options.
Laparoscopy is replacing open appendectomy for acute appendicitis, but new trials1-7 suggest that antibiotics can preclude surgery. Avoiding surgery may have attendant disadvantages, including longer hospitalization, prolonged outpatient therapy, and frequent failures. Many surgeons resist offering nonsurgical treatment, prompting further trials1-7 that have only reinforced current understanding of the advantages and disadvantages of this approach. Rather than conduct studies of whether surgery or antibiotic therapy is “better,” we asked possible patients what they would choose and why. Most respondents chose surgery, but some chose antibiotics.
How information is presented can influence respondent perceptions. If we had described rates of success for antibiotics in avoiding surgery rather than rates of failure in requiring surgery, more respondents might conceivably have chosen antibiotics. We chose this apparently negative language because this is the mode in which most surgeons generally obtain informed consent—discussing complication rates rather than the rates of avoiding complications. For instance, it seems unlikely that a surgeon describing surgical infection rates would say, “There is a 98% chance you won’t get a wound infection” instead of, “There is a 2% chance of wound infection.” Nevertheless, this mode of communication awaits future study.
Although most respondents preferred surgery (as did almost all surgeons), about one-tenth preferred antibiotics. This preference may be rational for some. Surgeons should offer antibiotics to patients with appendicitis while feeling comfortable explaining why they prefer surgery. Patients with difficult choices benefit from a cautious discussion of the physician’s own value-based preferences while noting that the patient may disagree.13 Although the law implies a standardized discussion of what “reasonable individuals” want to know, patients may want different information and make different value-based choices even with the same information.11,14-18 Physicians cannot always predict patients’ desires and thus should not assume that patients would reject nonsurgical treatment for appendicitis.19
Experience may have influenced choice. Surgeons and respondents with exposure to appendicitis chose surgery more frequently. Previous cases of appendicitis were most likely treated surgically, and respondents may have been reassured by satisfactory outcomes. Conversely, respondents whose friends or family had undergone surgery (not necessarily for appendicitis) were more likely to choose antibiotics, perhaps extrapolating from more difficult hospitalizations or convalescences after more serious disease to an expected painful or prolonged postappendectomy course. The increased likelihood of choosing antibiotics by respondents without family or friends who had been hospitalized could reflect an underestimation of the hospital stay associated with antibiotics.
Although both surgeons and nonsurgeons chose laparoscopic appendectomy much more frequently than open appendectomy, surgeons tended to be slightly more likely to choose open appendectomy than nonsurgeons. Most surgeons were within the age ranges in which one would expect laparoscopic experience, and surgeons between 40 and 49 years old were not substantially less likely to choose open appendectomy than were surgeons aged 60 years or older, for instance. Further study might explore this finding.
We hypothesized that risky health-and-safety behaviors might predict therapy choice. These particular risky behaviors were previously defined by Weber et al12 as a health-risk behavior subscale in a risk-behavior index. We used this subscale in our survey to determine whether such behaviors correlate with decision making in this particular personal medical decision. Because both antibiotics and surgery options entail some (but different) risks, decision making was likely influenced not only by the respondent’s likelihood of engaging in risky behavior (or a choice) but also by the respondent’s relative weighting of the severity of the risks posed by each choice. Overall health-risk scores did not predict treatment choice, but respondents were more likely to choose antibiotics for themselves if they were more likely to walk home alone at night in an unsafe area and for their children if they were less likely to consume 5 or more alcoholic drinks per day. This finding suggests that, first, medical decision making is linked to other behaviors and lifestyle choices and, second, all personal health-risk choices are not alike. Other medical decision making has been associated with the propensity to engage in specific health-risk behaviors, including alcohol abuse and smoking.20,21
If antibiotic therapy for acute appendicitis is reasonable but neither surgeons nor patients choose it, where should further research be directed? Additional trials6 seem unlikely to modify choices. Our results chart a different way forward. Respondents were most willing to choose antibiotics if short-term failure or long-term recurrence decreased. Traditional long courses of postoperative antibiotics are being shortened after definitive source control.22-24 Without source control, future research may more usefully emphasize selecting patient subsets who will benefit from current antibiotic protocols rather than attempt to shorten current hospitalization or outpatient antibiotic protocols while hoping that failure and recurrence rates will not worsen. For instance, given that patients presenting with appendicitis at night are sometimes managed by overnight antibiotics and morning appendectomy, the initial antibiotic response might select patients with lower failure rates. Conversely, pharmacologists might identify antibiotic protocols that reduce failure.
Our study has limitations. Hypothetical choices by healthy individuals might differ from choices of actual patients in distress. However, these hypothetical choices might be more rational, and the choices of healthy individuals interviewed in person in our second survey mirrored the results of the web-based survey. Different surgeons quote different complication and failure rates, but the outcomes in our scenarios were evidence based and seemed reasonable to surgeons in our community. Moreover, sensitivity analyses suggested that the overall conclusion that most people would choose surgery seems unlikely to have been changed by small variations in risk statistics.
In the web-based survey, we did not conduct formal testing to ensure that respondents fully understood all the information presented. However, the sensitivity analysis dialogue did demonstrate participant comprehension. We also could not ensure that participants would not respond twice from different computers, although there was no incentive to do so.
Finally, our convenience sample might have incompletely represented patients from underrepresented minority groups or other countries. Although our respondents represented diverse ages, occupations, and places of origin, more than 90% of them were non-Hispanic whites, limiting the racial or ethnic diversity of our sample. Respondents who self-identified as other than non-Hispanic white were more likely to choose antibiotics than those who self-identified as non-Hispanic white, although both groups were unlikely to choose antibiotics. The relatively small size of this subgroup precluded further exploration of this finding, but it suggests some cultural bias. White patients may be more likely than African American patients to choose surgery over nonoperative alternatives.25,26 Nevertheless, antibiotics alone were chosen very infrequently by any racial or ethnic group or any other subgroup. Ehlers et al6 reported that 47% of potential patients hypothetically would be willing to be randomized to surgical therapy or antibiotics therapy alone. This finding was higher than the fraction of respondents whom we found willing to choose antibiotics, perhaps because of the respondents' perceived value of participating in research or their belief that randomization would allocate them to their desired treatment. Nevertheless, even in our more hypothetical scenario, most respondents were unwilling to risk randomization to antibiotics.
Our results support surgeons’ prejudices that most patients with appendicitis want surgery and identify presumably rational patients who deserve the chance to choose antibiotics alone if they developed acute uncomplicated appendicitis. In a patient-centered world, patient values must trump surgeon values. Repeating similar trials may be unnecessary. Future efforts to improve the patient acceptability of antibiotics therapy for appendicitis should ideally focus on reducing the failure and recurrence rates of current protocols rather than shortening their duration.
Accepted for Publication: October 1, 2017.
Corresponding Author: Marc D. Basson, MD, PhD, MBA, Department of Surgery, University of North Dakota School of Medicine and Health Sciences, 1301 N Columbia Rd, Stop 9037, Grand Forks, ND 58202 (email@example.com).
Published Online: January 10, 2018. doi:10.1001/jamasurg.2017.5310
Author Contributions: Dr Basson had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Basson, Hanson.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Basson, Crosby.
Administrative, technical, or material support: Basson.
Study supervision: Basson.
Conflict of Interest Disclosures: None reported.
Create a personal account or sign in to: