This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Excessive use of antibiotics is very common worldwide, especially in rural China; various measures that have been used in curbing the problem have shown only marginal effects.
The objective of this study was to test an innovative intervention that provided just-in-time information and feedback (JITIF) to village doctors on care of common infectious diseases.
The information component of JITIF consisted of a set of theory or evidence-based ingredients, including operation guideline, public commitment, and takeaway information, whereas the feedback component tells each participating doctor about his or her performance scores and percentages of antibiotic prescriptions. These ingredients were incorporated together in a synergetic way via a Web-based aid. Evaluation of JITIF adopted a randomized controlled trial design involving 24 village clinics randomized into equal control and intervention arms. Measures used included changes between baseline and endpoint (1 year after baseline) in terms of: percentages of patients with symptomatic respiratory or gastrointestinal tract infections (RTIs or GTIs) being prescribed antibiotics, delivery of essential service procedures, and patients’ beliefs and knowledge about antibiotics and infection prevention. Two researchers worked as a group in collecting the data at each site clinic. One performed nonparticipative observation of the service process, while the other performed structured exit interviews about patients’ beliefs and knowledge. Data analysis comprised mainly of: (1) descriptive estimations of beliefs or knowledge, practice of indicative procedures, and use of antibiotics at baseline and endpoint for intervention and control groups and (2) chi-square tests for the differences between these groups.
A total of 1048 patients completed the evaluation, including 532 at baseline (intervention=269, control=263) and 516 at endpoint (intervention=262, control=254). Patients diagnosed with RTIs and GTIs accounted for 76.5% (407/532) and 23.5% (125/352), respectively, at baseline and 80.8% (417/532) and 19.2% (99/532) at endpoint. JITIF resulted in substantial improvement in delivery of essential service procedures (2.6%-24.8% at baseline on both arms and at endpoint on the control arm vs 88.5%-95.0% at endpoint on the intervention arm,
JITIF is effective in controlling antibiotics prescription at least in the short term and may provide a low-cost and sustainable solution to the widespread excessive use of antibiotics in rural China.
Antibiotics resistance (ABR) has become an urgent and worldwide public health problem. According to an independent review commissioned by UK Prime Minister David Cameron, around 700,000 people died in 2014 due to antimicrobial resistance and will reach 10 million lives a year by 2050, and a cumulative US $100 trillion of economic output are at risk due to drug-resistant infections if we do not find proactive solutions now [
A variety of measures have been used in improving antibiotics use (ABU) at primary settings. These include education of doctors and patients, structural changes, commitment letters, guidelines, prescription formularies, negative disease lists, shared decision-making (SDM), and others [
In response to the above challenges and others, this study aimed at developing and evaluating an innovative intervention, which provides the village doctors with just-in-time information and feedback (JITIF). Guided by proven theories and best practices, JITIF incorporates a number of cost-effective and sustainable approaches in a synergetic way in leveraging accountable ABU within routine primary care in rural China. Being designed as a potential package to be routinely used at rural primary care settings, JITIF covers all kinds of infections categorized as respiratory tract infections (RTIs), gastrointestinal tract infections (GTIs), urinary tract infections (UTIs), skin infections (SIs), and eye and dental infections (EDIs). Given that the prevalence of different infections varies greatly, we decided to test and promote JITIF in two stages. The first stage aims at testing its efficacy through a relatively small-scale randomized controlled trial focusing only on the two most common kinds of infections (ie, RTIs and GTIs), whereas the second leverages the experiences and findings from the first into a larger and more comprehensive trial covering all the aforementioned infections. This paper documents the main ingredients and mechanisms of the intervention package and preliminary findings about its efficacy from the first stage study.
As indicated by its title, the intervention comprises two components: information and feedback. The information component consisted of a set of theory and evidence-based ingredients, including operation guidelines, public commitment, and takeaway information. The operation guidelines divided routine care of symptomatic infection patients into 5 pragmatic steps (patient categorization, disease diagnosis, SDM, drug prescription, and patient education) and provided standard operation procedures (SOPs) for each of them.
The public commitment asked each of the participating village doctors in the intervention group to sign a letter of commitment and made the signed letter public by posting it on the walls of his or her clinic and printing it on the back of the patient takeaway information leaflet. The letter was first drafted by a researcher and then revised by all the intervention group doctors via a consensus session scheduled at the end of the project orientation training workshop. The finalized letter (see
The takeaway information (see
The feedback component of JITIF told each participating doctor about his or her performance scores (PSs) and percentages of prescribed ABU. The PSs for any individual doctor were based on the records of his or her management of symptomatic infection patients in the past 3 months and were rated by a panel of experts on care of infectious diseases according to a preset checklist. The percentages of prescribed ABU were also based on the same records and for the same time period but calculated automatically by the Web-based support system. Any PS (or percentage of ABU) for a given doctor was presented in red, yellow, and green, respectively, if it fell below (or above), within, and above (or below) the interquartile range of the same PS (or percentage of ABU) for all the participating doctors assessed in the same time period.
JITIF strived to deliver the abovementioned feedback and information at a time when the recipients were most ready to accept or act upon it. In reaching this end, it used a Web-based aid (WBA) consisting of 5 consecutive pages, each facilitated one of the 5 steps of routine service for infection patients mentioned above. A typical WBA page (see
Page 1 (see
Page 2 dealt with disease diagnosis. Its content varied depending on the category ticked in the previous page. Taking the example of RTIs, the SOP column of Page 2 proposed 3 substeps (SOP2-1 to SOP2-3). SOP2-1 examined common symptoms and signs of RTIs, with particular attention being paid on symptoms and symptom features helpful in distinguishing viral versus bacterial infections. SOP2-2 inquired previous formal and informal care for the current illness, especially ABU. SOP2-3 reached a clinical diagnosis of RTIs, for example, influenza, sore throat, and common cold. The performance column of Page 2 presented the doctor’s PSs in total and in terms of data completeness and/or accuracy for each of the 3 substeps included in the SOP column, whereas the reference column of Page 2 presented bulleted tips on soliciting information about RTI-related symptoms or signs, history, and distinguishing specific RTIs.
Page 3 leveraged SDM. It stressed adequate discussion between the attending doctor and attendee patient before prescribing any antibiotics. Its SOP column comprised 4 substeps. SOP3-1 predicted pathogen and occurrence and trajectory of patient’s symptoms. SOP3-2 asked the patient’s worries about the infection and expectations from the visit. SOP3-3 enumerated potential treatment options, including vacuum treatment (letting the patient go without prescribing any medicine), symptom relief treatment (carefully selected nonantibiotic alternative measures, for example, traditional food and physical therapies), backup antibiotic treatment and antibiotic treatment, and discussed benefits and side effects of each of them. SOP3-4 helped the patient in choosing one of the treatment options discussed. The performance column of Page 3 presented the doctor’s PS on this whole stage and on each of the 4 substeps mentioned above, whereas the reference column of Page 3 presented bulleted indications for using and not using antibiotics for the current type of infection (eg, RTIs, GTIs, and UTIs) and tips on practicing SDM.
Page 4 facilitated prescription if the treatment options chosen during the previous SDM contained antibiotic and/or nonantibiotic medications. Its SOP column provided prescription formularies, and its performance column provided percentage of antibiotic prescriptions by the attending doctor, as compared with that by his or her peers, for patients with different common infections, whereas its reference column provided bulleted rules for choosing medicines for the specific infection under concern (eg, common cold, influenza).
Page 5 promoted patient education. Its SOP column proposed 3 substeps. The content of SOP55-1 depended again on the treatment options chosen during the previous SDM. For options with antibiotic and nonantibiotic medications, it educated use of the medicines prescribed. For vacuum treatment, it proposed carefully selected alternative measures. For backup antibiotic treatment, it informed the patient when and on what indications he or she should come back to the doctor again. SOP5-2 counseled regarding the prevention of secondary infection to the patient’s relatives and future reinfection for the patient himself or herself. SOP55-3 developed and printed a tailored takeaway information sheet for the patient. The performance column of Page 5 showed PSs on this whole stage and on each of the 3 activities, whereas the reference column of Page 5 showed bulleted tips on counseling use of medications, backup treatment, and infection prevention.
Evaluation of JITIF adopted a randomized controlled trial design. The study settings included 12 intervention and 12 control villages in Anhui, a province located in east central China with a population of 68.6 million, of whom 57% live in rural areas. Per capita GDP and income in Anhui rank in the midrange (14 out of 31) among all provinces in China, and the social, cultural, and economic background in Anhui is representative of over 80% of the population in the nation. The study sites were determined via a 3-step clustered randomization. Step 1 divided all the 55 rural counties in Anhui province into north (17 counties), middle (16), and south (22). Step 2 randomly selected: (1) 4 counties from each of the regions, (2) 1 township from each of the selected counties, and (3) 2 administrative villages from each of the selected townships. Step 3 randomly assigned the 2 villages within each township into intervention and control arms. All the 24 village clinics in the selected villages agreed to participate. The 12 clinics on the intervention arm completed a baseline (from August 20 to September 30, 2015) and an endpoint (from August 22 to September 30, 2016) evaluation and implemented JITIF right after the baseline evaluation and throughout the whole study period, whereas the 12 clinics on the control arm completed only the baseline and endpoint evaluation. The study was not registered since randomization was applied only to clinics and not to patients.
Measures used in evaluating JITIF included changes between baseline and endpoint and between control and intervention groups in terms of: (1) percentages of patients with symptomatic RTIs or GTIs being prescribed with oral, intravenous, and injection antibiotics; (2) delivery of essential service procedures, including checking body signs, measuring temperature, performing auscultation, discussing nonantibiotic therapies, educating drug administration, counseling infection prevention, etc; and (3) patients’ knowledge about side effects of antibiotics and infection management and prevention. To enable examining compatibility between baseline versus endpoint and intervention versus control groups, the evaluation also collected data about patients’ age, sex, education, and clinical diagnosis.
Collection of the above data adopted nonparticipative observation using a structured form and face-to-face interview using a structured questionnaire. The observation form (see
The study took a combination of measures in ensuring data quality. These included: (1) training of field observers and interviewers on potential biases and measures avoiding them; (2) clarification of study purpose stressing that both positive and negative findings are of equal interest; and (3) feedback of rating-rerating discrepancies. Here, the feedback consisted of the following: (1) all the patient-doctor encounters were, after informed consent, audio-recorded; (2) a data quality supervisor randomly selected one case of the audio recordings from each of the site clinics every day, concealed all the labeling information, except a unique reference number of the recordings and then sent them to a third researcher; (3) the third researcher rated all applicable essential service procedures according to the audio recordings and using the same rating form as that used by the field observers and then sent the ratings back to the data quality supervisor; (4) the data quality supervisor calculated the differences between the ratings for the encounters with the same reference number but by different generators (one of the field observer and the third independent researcher); and (5) the data quality supervisor sent a short report about the average and observer-specific differences to each of the field observers every day during the data collection period.
The completed observation forms and questionnaires were double-entered into a database using EPI DATA (The EpiData Association. Version 3.1) and then analyzed using SPSS (IBM Corporation. Version 11) and Microsoft Excel (Microsoft Corporation. Version 2013). The data analysis consisted mainly of: (1) distribution of respondents and RTIs or GTIs by sociodemographics; (2) estimations of knowledge, delivery of indicative procedures, and use of antibiotics at baseline and endpoint and for intervention and control groups; and (3) chi-square power tests for differences in these estimations between different groups. For additional information about methods, please refer to
This study involved recruitment, intervention, and assessment of patients and village doctors. So it adhered to rigorous human subject protection principles and procedures. The study protocol had been reviewed and approved by the Biomedical Ethics Committee of Anhui Medical University before study commencement. Participation of villagers and village doctors was voluntary. Written informed consent was sought from all participants.
As shown in
Sociodemographic characteristics and symptomatic infections.
Participant group | Baseline | Endpoint | Overall |
|||||
Control (n=263), n (%) | Intervention (n=269), n (%) | Control (n=254), n (%) | Intervention (n=262), n (%) | |||||
.90 | .81 | .97 | ||||||
≤30 | 52 (19.8) | 53 (19.7) | 50 (19.7) | 54 (20.6) | ||||
31-40 | 51 (19.4) | 58 (21.6) | 49 (19.3) | 57 (21.8) | ||||
41-50 | 53 (20.2) | 59 (21.9) | 51 (20.1) | 55 (21.0) | ||||
51-60 | 52 (19.8) | 47 (17.5) | 49 (19.3) | 50 (19.1) | ||||
≥61 | 55 (20.8) | 52 (19.3) | 55 (21.6) | 46 (17.6) | ||||
.65 | .42 | .63 | ||||||
Male | 86 (32.7) | 83 (30.9) | 75 (29.5) | 86 (32.8) | ||||
Female | 177(67.3) | 186 (69.1) | 179 (70.5) | 176 (67.2) | ||||
.94 | .73 | .999 | ||||||
Illiteracy | 63 (24.0) | 60 (22.3) | 67 (26.4) | 58 (22.1) | ||||
Primary school | 72 (27.4) | 71 (26.4) | 61 (24.0) | 68 (26.0) | ||||
Middle school | 85 (32.3) | 92 (34.2) | 85 (33.5) | 91 (34.7) | ||||
High school | 43 (16.3) | 46 (17.1) | 41 (16.1) | 45(17.2) | ||||
.81 | .70 | .37 | ||||||
RTIsa |
200 (76.0) | 207 (77.0) | 207 (81.5) | 210 (80.2) | ||||
GTIsb |
63 (24.0) | 62 (23.0) | 47 (18.5) | 52 (19.8) |
aRTIs: respiratory tract infections.
bGTIs: gastrointestinal tract infections.
Flowchart of participant selection and follow-up.
Practices of essential service procedures at baseline and endpoint.
Procedure | Baseline | Endpoint | Overall |
|||||
Control, n (%) | Intervention, n (%) | Control, n (%) | Intervention, n (%) | |||||
Yes | 52 (19.8) | 54 (20.1) | .93 | 52 (20.5) | 237 (90.5) | <.001 | <.001 | |
No | 211 (80.2) | 215 (79.9) | 202 (79.5) | 25 (9.5) | ||||
Yes | 65 (24.7) | 59 (21.9) | .45 | 63 (24.8) | 236 (90.1) | <.001 | <.001 | |
No | 198 (75.3) | 210 (78.1) | 191 (75.2) | 26 (9.9) | ||||
Yes | 62 (23.6) | 57 (21.2) | .51 | 61 (24.0) | 238 (90.8) | <.001 | <.001 | |
No | 201 (76.4) | 212 (78.8) | 193 (76.0) | 24 (9.2) | ||||
Yes | 36 (13.7) | 37 (13.8) | .98 | 40 (15.7) | 232 (88.5) | <.001 | <.001 | |
No | 227 (86.3) | 232 (86.2) | 214 (84.3) | 30 (12.2) | ||||
Yes | 28 (10.6) | 27 (10.0) | .82 | 30 (11.8) | 242 (92.4) | <.001 | <.001 | |
No | 235 (89.4) | 242 (90.0) | 224 (88.2) | 20 (7.6) | ||||
Yes | 17 (6.5) | 15 (5.6) | .67 | 17 (6.7) | 242 (92.4) | <.001 | <.001 | |
No | 246 (93.5) | 254 (94.4) | 237 (93.3) | 20 (7.6) | ||||
Yes | 12 (4.6) | 16 (5.9) | .47 | 17 (6.7) | 247 (94.3) | <.001 | <.001 | |
No | 251 (95.4) | 253 (94.1) | 237 (93.3) | 15 (5.7) | ||||
Yes | 32 (12.2) | 31 (11.5) | .82 | 32 (12.6) | 249 (95.0) | <.001 | <.001 | |
No | 231 (87.8) | 238 (88.5) | 222 (87.4) | 13 (5.0) | ||||
Yes | 8 (3.0) | 7 (2.6) | .76 | 7 (2.8) | 248 (94.7) | <.001 | <.001 | |
No | 255 (97.0) | 262 (97.4) | 247 (97.2) | 14 (5.3) | ||||
Yes | 9 (3.4) | 8 (3.0) | .77 | 8 (3.1) | 241 (92.0) | <.001 | <.001 | |
No | 254 (96.6) | 261 (97.0) | 246 (96.9) | 21 (8.0) |
aRTIs: respiratory tract infections.
bGTIs: gastrointestinal tract infections.
Changes in patients’ beliefs and knowledge about antibiotics and infections.
Questionnaire item | Baseline | Endpoint | Overall |
|||||
Control, n (%) | Intervention, n (%) | Control, n (%) | Intervention, n (%) | |||||
Yes or Not clear | 189 (71.9) | 192 (71.4) | .90 | 182 (71.7) | 156 (59.5) | .004 | .004 | |
No | 74 (27.2) | 77 (28.6) | 72 (28.3) | 106 (40.5) | ||||
Yes or Not clear | 223 (84.8) | 229 (85.1) | .91 | 213 (83.9) | 204 (77.9) | .08 | .03 | |
No | 40 (15.2) | 40 (14.9) | 41 (16.1) | 58 (22.1) | ||||
Yes or Not clear | 184 (70.0) | 188 (69.9) | .96 | 171 (67.3) | 148 (56.5) | .01 | .001 | |
No | 79 (30.0) | 81 (30.1) | 83 (32.7) | 114 (43.5) | ||||
Yes or Not clear | 187 (71.1) | 192 (71.4) | .95 | 181 (71.3) | 157 (59.9) | .01 | .005 | |
No | 76 (28.9) | 77 (28.6) | 73 (28.7) | 105 (40.1) | ||||
Yes or Not clear | 175 (66.5) | 172 (63.9) | .53 | 153 (60.2) | 98 (37.4) | <.001 | <.001 | |
No | 88 (33.5) | 97 (36.1) | 101 (39.8) | 164 (62.6) | ||||
Yes or Not clear | 231 (87.8) | 238 (88.5) | .82 | 221 (87.0) | 210 (80.2) | .04 | .008 | |
No | 32 (12.2) | 31 (11.5) | 33 (13.0) | 52 (19.8) | ||||
Yes or Not clear | 166 (63.1) | 172 (63.9) | .84 | 158 (62.2) | 130 (49.6) | .004 | .001 | |
No | 97 (36.9) | 97 (36.1) | 96 (37.8) | 132 (50.4) | ||||
Yes | 92 (35.0) | 96 (35.7) | .87 | 91 (35.8) | 193 (73.7) | .001 | <.001 | |
No or Not clear | 171 (65.0) | 173 (64.3) | 163 (64.2) | 69 (26.3) | ||||
Yes | 80 (40.0) | 81 (39.1) | .86 | 88 (42.5) | 140 (66.7) | <.001 | <.001 | |
No or Not clear | 119 (60.0) | 126 (60.9) | 119 (57.5) | 70 (33.3) | ||||
Yes | 30 (47.6) | 29 (46.8) | .93 | 22 (46.8) | 36 (69.2) | .02 | .02 | |
No or Not clear | 33 (52.4) | 33 (53.2) | 25 (53.2) | 16 (30.8) |
aRTIs: respiratory tract infections.
bGTIs: gastrointestinal tract infections.
Antibiotics prescription by baseline, endpoint, control, and intervention groups.
Antibiotics prescription | Baseline | Endpoint | Overall |
|||||
Control, n (%) | Intervention, n (%) | Control, n (%) | Intervention, n (%) | |||||
Oral | 97 (66.0) | 122 (65.6) | .94 | 92 (60.5) | 85 (42.3) | .001 | <.001 | |
Intravenous | 89 (53.0) | 95 (54.0) | .89 | 85 (51.2) | 71 (38.4) | .02 | .003 | |
Any | 159 (90.3) | 176 (87.1) | .33 | 156 (89.7) | 135 (64.3) | <.001 | <.001 | |
Oral | 34 (68.0) | 32 (64.0) | .67 | 25 (62.5) | 6 (12.8) | <.001 | <.001 | |
Intravenous | 30 (60.0) | 34 (58.6) | .88 | 21 (60.0) | 16 (36.4) | .04 | .03 | |
Any | 50 (96.2) | 54 (94.7) | .72 | 36 (90.0) | 22 (52.4) | <.001 | <.001 | |
Oral | 131 (66.5) | 154 (65.3) | .79 | 117 (60.9) | 91 (36.7) | <.001 | <.001 | |
Intravenous | 119 (54.6) | 129 (55.1) | .91 | 106 (52.7) | 87 (38.0) | .002 | <.001 | |
Any | 209 (91.7) | 230 (88.8) | .29 | 192 (89.7) | 157 (62.3) | <.001 | <.001 |
aRTIs: respiratory tract infections.
bGTIs: gastrointestinal tract infections.
As mentioned earlier in the Introduction, many studies have been performed to determine the effectiveness of different types of interventions in promoting a more rational use of antibiotics at primary care settings, and multifaceted interventions have generally been more successful, especially if they employ SDM, involve peers, and benefit the practice as a whole [
Perhaps, the primary reason underlying the efficacy of this study may be that JITIF incorporates a number of theory- or evidence-based approaches in a synergetic way. The public commitment, for example, originated from an alternative behavior model holds that: (1) people place a high value on consistency and follow through with their public commitment to avoid disapproval by their peers and (2) publicly committing to a behavior prompts people to later justify that behavior and identify the behavior with their self-image, which may enhance personal dedication to performing that behavior [
The study also revealed useful information for better understanding routine health care for patients with symptomatic RTIs or GTIs at primary care settings in rural Anhui, China. Each of the procedures observed in this study (
This study has both strength and weakness. Its strength originates from: (1) a packaged intervention consisting of cost-effective and sustainable approaches incorporated in a synergetic way and (2) an efficacy evaluation using a randomized controlled trial design that involved both baseline versus endpoint and control versus intervention comparisons. Perhaps, the biggest concern of the study may be observation-induced interferences on the practice behaviors. When being observed, the doctors may be more compliant to authorized guidelines [
In conclusion, excessive use of antibiotics was very prevalent, and most essential service procedures for patients with symptomatic infections were not commonly practiced at primary care settings in rural Anhui, China. JITIF was effective in reducing antibiotic use and improving service procedures.
Commitment letter.
Sample patient takeaway information.
Sample screenshot of Web-based aid.
Checklist for observation of essential service procedures.
Additional information on web-based aid for just-in-time information and feedback (JITIF-WBA).
CONSORT E-HEALTH checklist (V 1.6.1).
antibiotic resistance
antibiotic use
eye and dental infection
gastrointestinal tract infection
just-in-time information and feedback
performance score
respiratory tract infection
shared decision-making
skin infection
standard operation procedure
urinary tract infection
Web-based aid
Development of the primitive project protocol was supported by the China-UK Prosperity Fund (grant number: PPY CHN 1590/15SS19), whereas implementation of the study by the Science Foundation of China (grant number: 81661138001).
XS conceived this project, facilitated the protocol and instrument development, and drafted this manuscript. ML and RF led field data collection. J Chai programmed the website. J Cheng supervised field data collection. MX, MD, and TJ analyzed the data. DW provided expertise for the overall design of the study and finalized the manuscript.
None declared.