Is fasting safe? A chart review of adverse events during medically supervised, water-only fasting

TNHC (www.​truenorthhealth.​com) is an integrative medical facility in Santa Rosa, CA, USA that offers a residential health education program specializing in water-only fasting and dietary intervention. Patients in this study had access to medical, naturopathic, and chiropractic doctors. Private rooms, 24-h medical supervision, and daily educational activities were also provided.

The detailed water-only fasting protocol used in this study is within the standards established by the International Association of Hygienic Physicians (Additional file 1: Supplementary Methods) [40]. Briefly, patients electing to water-only fast at TNHC routinely underwent a comprehensive physical, neurological, and psychological examination including medical history, urinalysis, complete blood count with differentials, and comprehensive metabolic panel. Patients with contraindications or taking medications that could not be discontinued, with the exception of thyroid medications which can be continued at a reduced concentration, were not admitted for water-only fasting.

Patients were instructed to eat a diet of fresh raw fruits and vegetables and steamed starchy vegetables for at least 2 days before fasting. During the fast, patients remained on site, drank a minimum of 40 ounces of distilled water per day, and minimized physical activity. Medical staff monitored fasting patients twice daily and recorded vital signs and symptoms into chart notes. Urinalysis and the aforementioned blood tests were repeated weekly or as directed by a clinician. Water-only fasts were discontinued when symptoms stabilized, the patient requested termination of the fast, or the clinician deemed it necessary for medical reasons.

Following the fast, patients refed for a period of time lasting half of the fast length. Standard refeeding consisted of five phases (1 phase for each 7–10 days of water-only fasting) (Supplementary Methods), beginning with juice and followed by gradual introduction of solid plant foods (free from added sugar, oil, and salt) (Additional file 1: Supplementary Methods). Once patients were eating solid foods, the gradual reintroduction of moderate exercise was allowed. During the refeeding process, patients were offered twice-daily monitoring by clinicians.

Electronic charts from May, 2006 to December 31, 2011 were extracted into a single relational database (MS Access). The data collected included age, gender, chief complaints, treatment status (i.e., water, juice, broth, or food), and clinical chart notes. Chief complaints data were derived from self-reports and clinical diagnoses.

The study included visits by patients who water-only fasted for at least two consecutive days and refed for at least half of the fast length. The study excluded visits by patients who fasted prior to Jan. 1, 2007 because of incomplete data records, were under twenty-one years old, water-only fasted for one day or less because this length is defined as an intermittent fast and generally does not require medical supervision, [41] interrupted the water-only fast with any amount of vegetable broth or juice and/or fruit juice, had supervised refeeding less than half of the fast length, or if there were insufficient data to determine treatment status on a given day (Fig. 1). Out of 2539 visits, 768 were included in our study. Of the selected visits, the analysis dataset included visit days from the day fasting began to day the refeeding period (1/2 length of fast post-fast) ended. For fasts of an odd number of days, the refeeding period was rounded up to the nearest integer (e.g., if a visit had 3 days of fasting, then analysis data included 2 days of refeeding). There was no information about the condition of the patient other than treatment status for 5.2% (496/9570) of total visit days (312 during fasting/184 refeeding).

Data on AEs were collected from clinical chart notes of self-reported symptoms, clinical and diagnostic findings, and medical management of symptoms. Laboratory values were not systematically included due to data incompleteness. The CTCAE (v4.03) scale was used to grade AE severity: AE1 – mild, AE2 – moderate, AE3 – severe, AE4 – life threatening or disabling, and AE5 – death. AEs were characterized using the MedDRA (v12.1) terminology according to the system organ class (SOC), which uses the highest level term (HLT), including anatomical or physiological system, etiology, and/or purpose, and the lowest level term (LLT). Pain-related AEs, reported on a visual-analog scale (VAS) from 0 to 10, were reported as follows: AE1 = VAS 1–4, AE2 = VAS 5–7, and AE3 = VAS 8–10 [42, 43]. A serious adverse event (SAE) was determined based on the definition used by the Department of Health and Human Services [44]. A codification guide was created to ensure that there was consistent codification of AEs. A single trained clinician (J.S.F) identified, reviewed, and classified AEs, according to the above criteria, for each daily entry in the electronic chart note. A user interface within the relational database (MS Access) was used for direct data entry of the codified data. One week after the conclusion of data abstraction, intra-rater reliability was assessed based on a random sample of daily logs. The dataset is available from the Dryad repository, https://​doi.​org/​10.​5061/​dryad.​6cg6j.

Inter-rater reliability was assessed using a blind, random sample of daily logs (n = 645, 7%). There was 80.5% agreement between MedDRA codes identified in the original coding and secondary assessment. Of these, the weighted kappa statistic (squared weights) measuring agreement of the AE grade was 0.92 [45].

Visits were stratified by water-only fast length: 2–7 (short), 8–14 (medium), 15–21 (long), or 22+ (extended) days. These categories were chosen to represent lengths of typical fasts. Within visits, days were grouped by fasting and refeeding status. Summary statistics of the study population at the patient and visit levels were tabulated. Chief complaints were categorized by SOC when possible. For each visit, the highest grade AE (HGAE) experienced (of any MedDRA term) and the total AEs were calculated for the fasting, refeeding, and entire protocol periods. Tallies and percentages of HGAE within each of these periods were computed in total and by the fasting length strata. Spearman’s correlation between fasting length strata and HGAE was computed using integer scores for the fasting length strata. For each MedDRA term, counts and percentages of visits with at least one AE of a term were computed for each AE grade and for all AE grades combined. Similar counts and percentages were computed for individual AEs without taking visits into account. We report MedDRA terms that were experienced in more than 10%. Association between a visit’s HGAE and age was assessed graphically and measured with Spearman’s correlation [46]. An association between HGAE and gender was tested using a two-sided Fisher’s exact test. The paired difference in HGAE during fasting versus refeeding was tested using a paired t-test. Kaplan-Meier estimates of cumulative incidence (CI) were computed for each category of fasting length for the following AE categories: any AE, any AE ≥ 2, and any AE ≥ 3 [47]. All AEs of grade 3 or higher were carefully reviewed (A.C.G., T.R.M, J.S.F.), and SAEs were further described by brief narrative. Analyses were performed in R 3.1 using the survival, [48, 49] irr, [45] ggplot2, [50] dplyr, [51] and htmlTable packages [52].

We analyzed charts from 768 visits of patients who water-only fasted at TNHC from 2007 to 2011 for at least 2 consecutive days followed by a refeeding period equal to half of the fast length. There were a total of 652 patients, 409 (63%) were female, and the median (interquartile range, IQR) age at first visit was 55 (17.25) years old. Of these, 54 patients fasted 2 times and 23 fasted 3 or more times during the study period.

There were 768 visits, comprising 9570 visit days and 6265 fasting days. The median fasting length was 7 (IQR 6) days; the shortest fast was 2 days and longest fast was 41 days. Of these, there were 446 short (2–7 fasting days, median [IQR] = 5 [3] days), 238 medium (8–14 fasting days, median [IQR] = 10 [3] days), 64 long (14–21 fasting days, median [IQR] = 18 [4] days), and 20 extended (22 or more fasting days, median [IQR] = 27.5 [6.2] days) visits. Table 1 describes chief complaint categories as total counts and as a percentage of total visits. A patient could have had more than one chief complaint per visit. Quality of life, including prevention and fatigue, was the primary reason patients visited TNHC (n = 384, 50%). Other major chief complaint categories included the cardiovascular (n = 193, 25.1%), musculoskeletal (n = 147, 19.1%), gastrointestinal (n = 122, 15.9%), and endocrine (n = 107, 13.9%) systems. Following prevention (n = 358, 46.7%), hypertension (n = 152, 19.7%) was the largest chief complaint category.

AEs were classified on a graded scale of 1–5 according to CTCAE (v4.03) and MedDRA (v12.1) criteria (see Methods). HGAE experienced during the fasting, refeeding, and entire protocol period of each visit is presented in Table 2. During the entire protocol period (i.e., fasting and refeeding) for all visit lengths, the HGAE experienced in the majority of visits (n = 555, 72.3%) was grade 2 or lower and in 212 visits (26.6%) it was grade 3. There was a weak positive correlation between fast and refeed length and HGAE (Table 2). There was 1 patient with a grade 4 AE during a medium visit. There were no deaths (AE5) during any visits. There was a weak positive correlation between HGAE and fast duration (Spearman’s ρ = 0.28, p = < 0.001) and a weak positive correlation between HGAE and age (Spearman’s ρ = 0.11, p = 0.002) (Additional file 1: Figure S1a, b). There was no correlation between HGAE and gender (p = 0.628) (Additional file 1: Figure S1c). During refeeding, the mean HGAE was 0.56 less than during fasting (paired t = − 16.28, p = < 0.001).

Kaplan-Meier CI plots for AEs during fasting and during refeeding are shown in Fig. 2. At day 5 of fasting, the estimated cumulative incidence (ECI) for experiencing any AE was 0.90 (95% CI: 0.88, 0.92) (Fig. 2a). For experiencing any AE ≥ grade 2, the ECI was 0.54 (0.50, 0.58) at day 5 and 0.69 (0.65, 0.73) at day 10 (Fig. 2b). For any AE ≥ grade 3, the ECI was 0.20 (0.17, 0.23) at day 5, 0.28 (0.24, 0.32) at day 10, and 0.32 (0.27, 0.37) at day 15 (Fig. 2c).

MedDRA classification and HGAE that occur in greater than 10% of visits are presented in Table 3. The results indicate visits during which the subject experienced a given AE classification at least one time; subjects could have experienced more than one AE class during a single visit. During the majority of visits, subjects experienced AEs that were classified as fatigue (n = 370, 48.2%), nausea (n = 247, 32.2%), insomnia (n = 257, 33.5%), headache (n = 231, 30.1%), hypertension (n = 224, 29.2%), presyncope (n = 217, 28.3%), dyspepsia (n = 199, 25.9%), and back pain (n = 197, 25.7%). Presyncope (n = 212) and hypertension (n = 77) occurred in the majority of visits in which grade 2 and grade 3 was the HGAE experienced, respectively. Of the 77 visits with a grade 3 hypertension AE, 75 had hypertension as a chief complaint upon arrival.

MedDRA classification and AE grade of the total number of AEs are presented in (Additional file 1: Table S1). There were a total of 5961 AEs. Of these, 75.3% were AE1, 19.6% were AE2, 5.0% were AE3, and 0.0003% were AE4. The most common events were the same as those that occurred in the majority of visits in Table 3.

There were two (0.002% of visits) SAEs that required hospitalization. One was a grade 3 dehydration event that occurred on fasting day 3 in a 73-year-old, male patient. The patient developed headache, fever, and increased blood pressure. He was taken to an emergency facility, where he received antibiotics for potential upper respiratory infection and intravenous (IV) electrolytes for dehydration and was kept under inpatient observation for 3 days. The patient fully recovered and returned to TNHC. The other was a grade 4 hyponatremia event that occurred on fasting day 9 in a 70-year-old, male patient. The patient developed difficulty with speech and was immediately transported to an emergency facility by ambulance. Upon evaluation at the emergency facility, the patient was diagnosed with hyponatremia, administered IV electrolytes, and kept under inpatient observation for 4 days. The patient fully recovered but did not return to TNHC.