Alcohol-related liver disease (ALD), including cirrhosis-related death and hepatocellular carcinoma, is a leading cause of liver-related death [1,2]. In the Asia-Pacific region, alcohol has become the third most common cause of liver-related deaths after hepatitis B and hepatitis C [3]. Consequently, liver disease etiology has been changing from viral hepatitis to non-viral liver diseases [4-7].

As one part of treatment strategies, abstinence from alcohol is strongly recommended for patients with alcoholic cirrhosis or alcoholic hepatitis [8]. In primary care settings, “screening and brief intervention” are effective to reduce drinking in the general population [9]. Recent guidelines have also recommended this screening and brief intervention approach for patients with ALD [8,10]. An Alcohol Use Disorders Inventory Test (AUDIT) score greater than 8 is predictive of harmful or hazardous alcohol use. A score greater than 20 is suggestive of moderate-to-severe alcohol use disorder (AUD) [8,10]. Alcohol screening questionnaires including CAGE [11,12] and the AUDIT-C (3 questions related to alcohol consumption from the full questionnaire) [13-15] in an outpatient setting are useful to predict subsequent hospitalization for gastrointestinal conditions [16]. Moreover, the usefulness of the AUDIT-C to identify risks of alcohol relapse after liver transplantation has also been shown [17].

For patients with moderate-to-severe AUD, a brief intervention is not sufficient. Referral to treatment for AUD by an addiction specialist is ideal [8,10]. However, among high-risk drinkers, the proportion of patients receiving treatment for liver dysfunction or alcoholism is low. A survey revealed that only 0.8% of risky drinkers wish to stop drinking and that only 5.1% of risky drinkers are undergoing treatment for liver dysfunction [18]. These reports suggest that many patients with ALD and AUD refuse to reduce drinking or to accept addiction counseling. Considering this background, the key issue with AUD treatment for patients with ALD is reduction of patients’ drinking by hepatologists, even without the support of addiction specialists.

Digital tools including smartphone apps might be used as feedback and planning support, educational tools, or other support in order to improve health care [19].

Feedback is a core service supported by most systems. Popular feedback items provided to users are cost and calorie equivalents to a person’s alcoholic beverage consumption, as well as drinking-related health risks [19,20]. Scott and colleagues [21] reported a project to assess granular self-monitoring feedback using ecological momentary assessments and immediate recovery support through ecological momentary interventions.

Educational support provides knowledge related to diseases (alcohol addiction) and risks associated with alcohol. It enables the encouragement of self-motivated alcohol reduction. Drinkaware [22] proposes the following educational factors: assessment and feedback, high-risk places for drinking, choosing and using supportive people for change, cravings and their management, problem-solving skills, communication, refusal to drink skills, and fun nondrinking activities.

Emergency support is useful in managing expected or out-of-control situations. The Addiction-Comprehensive Health Enhancement Support System (A-CHESS) [23] provides emergency support called the panic button. If the patient presses the panic button, then the system sends an emergency message to 2 contact persons who have been prepared in advance.

Most apps have successfully demonstrated effectiveness. One key issue is safety. For example, Glowacki et al [24] reported that a mismatched text message sometimes exacerbates unhealthy behavior.

Another issue is that these approaches have been examined in settings of alcohol dependence but not in settings of ALD. Reported smartphone apps are used mainly by patients [20-26]; physicians do not use them.

A smartphone app is presumably able to provide rich patient information to hepatologists, leading to effective feedback. To assess patient information, a physician (hepatologist) can review the following 2 types of information from a patient's input: (1) alcohol consumption that indicates the patient's drinking status and (2) a patient narrative that can imply the patient's motivation. Using that information, hepatologists can give suitable feedback to patients.

The objectives of this study were to assess the use of a smartphone app as a journaling tool and a source of feedback based on self-report for patients with ALD, specifically its effects on (1) abstinence from alcohol and (2) improvement in biomarkers of alcohol use and liver function tests.

The target hospital was a main hospital in the Osaka area of Japan. Approximately 200 patients with liver disease are seen in the outpatient department each week, including about 10 cases referred from general practitioners. During August 2021 to May 2022, we recruited patients with ALD who were seen either in the hospital or an outpatient clinic. When the patient’s liver disease was regarded as caused by harmful alcohol intake, patients were given educational booklets that included information about health risks and harmful alcohol use. Abstinence from alcohol was recommended by hepatologists (NY, MM, KS), especially for patients with liver cirrhosis or hepatocellular carcinoma. This was advice with no implied nor actual penalty. When AUD was suspected based on the AUDIT score [13], psychiatric outpatient care was provided with consent, but no recovery program was routinely provided at our hospital. When an ALD patient was admitted to our hospital for any liver-related condition, a psychiatric liaison team round was scheduled by agreement.

No patient who had been abstinent for more than 1 year or who had refused to reduce or abstain from alcohol intake was invited to participate in the study. During the period, 23 patients were invited, 14 of whom provided written informed consent. The patient flow is portrayed in Figure 1.

Outpatient appointments were set at week 4, week 8, and week 24. However, changes to appointments were allowed 1 week before and after the date that had been initially set.

Patient physical data and clinical characteristics were obtained at the time of entry into the study. Laboratory data, including aspartate aminotransferase, alanine aminotransferase, gamma glutamyl transferase (GGT), total bilirubin, albumin, blood cell count, prothrombin international normalized ratio, and serum percentage of carbohydrate-deficient transferrin to transferrin (%CDT) level, were measured at baseline, week 4, week 8 and week 24. The level of serum %CDT was analyzed using direct immunonephelometric assay (N Latex CDT; Siemens Healthcare Diagnostics). A combined index based on GGT and CDT measurements (GGT-CDT) has been shown to improve sensitivity to detect heavy drinkers and was calculated as [27] GGT-CDT = 0.8 × ln(GGT) + 1.3 × ln(%CDT).

At each follow-up visit (week 4 and week 8), the treating hepatologist (NY, MM, and KS) provided self-report–based feedback in addition to regular outpatient practice. The interventions in this study are presented in Table 1. The hepatologist referred to the “physician’s view” to review whether drinking status was reported, how much ethanol was consumed on average, and what type of input was made. An example of using a physician’s view in an outpatient setting is the following: A patient and a clinician viewed the graph of alcohol consumption together, thereby enabling the patient to confirm the amount and frequency of alcohol consumption. However, we used no specific feedback protocol. The feedback manner differed among hepatologists because no special training for behavioral health was executed.

There were 2 developed systems: (1) a client smartphone journaling app for patients and (2) a server app for physicians’ use.

Normally or near-normally distributed variables are reported as means and SDs. Equality between groups was compared using the Student t test. The results obtained before and after the intervention were compared using paired t tests. Non-normally distributed continuous data are reported as medians and ranges. They were compared using Wilcoxon rank-sum tests. Categorical variables were compared using chi-square tests or Fisher exact tests when appropriate. Correlations between self-reported average daily ethanol consumption and biomarkers of alcohol use were analyzed using Pearson correlation coefficient tests. All analyses were conducted using software (STATA ver. 15; Stata Corp). P values less than .05 were considered significant.

This study, which was evaluated and approved by the institutional ethics committee (No.2020252), was performed in compliance with the Declaration of Helsinki.

This pilot study evaluated the use of a smartphone journaling app by patients with ALD. About 60% of candidate patients were evaluated in this study. Of them, 71% completed 8 weeks of journaling. One-half of the patients who continued diary input achieved 8 weeks of complete abstinence.

With ALD on the rise, efforts for harm reduction and alcohol consumption reduction are attracting attention [28]. For patients with both ALD and AUD, screening and brief interventions and referral to treatment are useful. Subsequent pharmacotherapy and psychiatric intervention such as cognitive therapy and motivation enhancement therapy might also be useful [10,28]. However, the percentage of patients actually receiving treatment is probably small [13,29]. Patients might visit a hospital and might receive a diagnosis of ALD only after symptoms such as jaundice appear. Some patients are resistant to intervention by psychiatric and multidisciplinary teams because they are not properly diagnosed with AUD or because they do not accept a diagnosis of AUD.

Biomarkers are convenient for assessing alcohol consumption in an outpatient setting [30,31]. Routinely available laboratory tests such as mean corpuscular volume and GGT can be done easily, but they do not show greater amounts of alcohol consumption directly. Blood-urine alcohol concentration has a short half-life. For that reason, it might not reflect the amount of alcohol consumed during outpatient evaluations. The %CDT has been used as a biomarker to detect heavy (>50 g) alcohol consumption within the prior several weeks [10,30-32]. Suzuki et al [32] reported a cut-off value for excessive drinkers (men ≥60 g/day, women ≥40 g/day) of 1.9% (sensitivity=0.77, specificity=0.77). Reportedly, the combination of GGT and %CDT (GGT-CDT) shows higher sensitivity for detection of heavy drinking and higher prediction accuracy for detection of excessive alcohol consumption [27,33]. Hietala et al [27] reported the cut-off value for heavy drinkers as 4.18 for men and as 3.81 for women using a derived formula (sensitivity=0.90, specificity=0.98). However, the use of %CDT for patients with liver disease remains limited because of its low sensitivity and low specificity [30]. Elevated %CDT values have been noted frequently in liver transplant candidates with end-stage liver disease [34]. Currently, other biomarkers such as ethyl glucuronide and phosphatidylethanol are used widely [35].

For this study, liver function and %CDT were evaluated because other alcohol consumption–specific markers have not been commercially available. Although the number of patients was too low to evaluate the diagnostic performance of %CDT, this measure can reflect changes related to alcohol consumption for time series monitoring of individual cases. However, we found no correlation between self-reported ethanol consumption and %CDT. The sensitivity of %CDT was low probably because of the large number of patients with advanced cirrhosis.

The usefulness of diaries for recording alcohol consumption was demonstrated, but their potential as an adjunctive test for judging treatment effects was also suggested by our results.

Nevertheless, the tendencies found in diary entries suggest that input rates were high among abstinent patients, whereas the rates were low among continuing drinkers, also indicating a long lag separating the date of entry and the date of diary completion. Considering the resistance of patients to having their drinking habits known by their treating physicians, some patients might refuse direct testing for alcohol consumption as well.

One-half of the patients who completed the study achieved 8 weeks of self-reported abstinence. By checking the physician's view of the server app, the daily drinking status can be noted at a glance. The alcohol consumption amount was graphed, making it possible to ascertain the self-reported amount of alcohol consumed. Most patients in this study had liver cirrhosis. Of the 5 patients who abstained from alcohol, 4 had severe Child class C cirrhosis, suggesting that more severe conditions might be a factor facilitating abstinence. Other factors such as the absence of a comorbid psychiatric disorder and having a job were also notable.

For a short period of time, a trend toward improved nutritional status (albumin) was apparent as a result of abstinence by patients using the smartphone journaling app. Other indicators were not found to be significant in this study. However, the fact that patients' proactive approaches to abstinence behavior engender improved liver function can motivate them to seek and benefit from future treatment.

One benefit of using a smartphone app is the possibility of identifying when the patient entered a record. For paper-based diaries, it is difficult to ascertain when the patient makes an entry. In this study, the results of analyzing input record date lags in the data indicated that patients who were sober tended to input data into the smartphone journaling app almost every day. By contrast, nonabstaining patients tended to enter many records together at one time, according to the date of their medical visit. In such cases, the input record date lags tended to be greater in the fourth and eighth weeks of the study. Based on these considerations, a certain number of patients probably used the smartphone journaling app with some awareness of the date of their medical visit. Therefore, providing automatic feedback through the smartphone app is necessary to raise awareness and motivation. Alternatively, an input time lag might provide a hint that a patient needs more care.

This study has several limitations. The smartphone journaling app used for this study was set up so that a one-line diary could not be saved without entering a single word, irrespective of whether the user had been drinking. Results from a questionnaire completed after using the app indicated that “entering daily diary entries is difficult.” Possible future improvements to the app described herein include allowing free-form diary entries and allowing a stamp function that can help users simply express their mood on a given day.

Another limitation is this current study had no control arm participants who had received treatment (4 weeks and 8 weeks) without the smartphone app. We assume that such control cases would show no significant change in the short term (4 weeks or 8 weeks). However, a larger scale and double-arm, randomized controlled trial is highly desirable.

Nonetheless, this pilot study showed positive results among patients with ALD who were recommended to reduce or quit drinking. Among multiple functions provided by the smartphone app (journaling, physician’s view-based feedback), precisely which function was the main trigger demands further study.

In conclusion, the use of an alcohol use journaling app in combination with hepatologists’ follow-up was associated with reduction of daily alcohol intake and improvement of liver function test results. The positional gap separating patients and physicians was narrowed by sharing the journaling app. Further studies are expected to elucidate the use of digital tools in an ambulatory care setting.

The data sets analyzed during the current study are not publicly available for reasons of patient privacy, but they are available from the corresponding author on reasonable request.