Absconding During Inpatient Care from a Tertiary Psychiatric Hospital: A Comparative Study.

BACKGROUND: Absconding from psychiatric hospitals is of great concern for patients and caregivers. Absconding affects not only the treatment and safety of these patients but also patient's caregivers and the community. Further investigation is needed to examine the pattern of this event and the characteristics of patients who abscond. Hence, our study was aimed to examine the sociodemographic and clinical profiles of inpatients who absconded from a psychiatric hospital in five years and to compare them with matched controls.
METHODS: A retrospective chart review of inpatients who absconded and matched control inpatients during the specified period of five years from January 2014 to December 2018 was done at a psychiatric hospital. Each control was matched with a corresponding absconding case on the following order: (a) admission ward, (b) admission period, (c) diagnosis, and (d) age.
Results: Among 20,052 adult admissions during the specified period, 38 patients absconded, with a rate of 1.8 per 1,000 admissions. Most of them were male, from a younger age group, diagnosed with schizophrenia or mood disorder, and having comorbid substance use disorder, irritable affect, impaired judgment, and absent insight. Most of the events occurred within the first two weeks of admission. About 11% of them had a history of prior absconding from the hospital.
CONCLUSION: Knowledge about the associated sociodemographic and clinical profile would help clinicians and mental health care professionals to prevent absconding. Further risk assessment using a patient's profile would help to reduce absconding events from psychiatric hospitals in the future.

Key Messages: Most of the inpatients who absconded were male, from a younger age group, diagnosed with schizophrenia or mood disorder, with comorbid substance use disorder and had irritable affect, impaired judgment, and absent insight. Most incidents of absconding occurred during the first two weeks following admission.

Absconding of inpatients from psychiatric hospitals is of great concern for patients and caregivers. As a result of absconding, treatment and safety of these patients is affected, and it may also affect caregivers of the patient and the community a patient belongs to. Further investigation is needed to examine the pattern of this event and the characteristics of patients who abscond.[1, 2] Absconding is associated with an increased risk for suicide, self-harm, homicide, and becoming “missing” from society. Rates of absconding from psychiatric care are different across different parts of the world.[3] A few studies have examined the characteristics of patients who abscond from psychiatric facilities, for various sociodemographic and clinical reasons.[1, 4]–[18] An “absconder” has been defined as “any patient who left his hospital without medical advice and failed to return by midnight, and also any compulsory patient who failed to return from leave after his authorized period of leave had expired.”[8] Broadly, absconding has been defined by researchers as leaving hospital without permission.[4, 5, 9, 12] Often, studies have found that those who abscond are young, male, diagnosed with schizophrenia, and have a short duration of stay and a previous history of absconding. Some studies have found significantly lower rates of absconding in general hospitals than psychiatric hospitals or psychiatric wards.[14-19] This might be due to differences in inpatients’ clinical profiles and hospital environment.[15] A recent review revealed that patients abscond to find relief, and to regain power and control over their lives. Absconding has been viewed as a means of seeking freedom.[20] A few studies from India also have investigated this.[19, 21–24] A literature review by Stewart and Bowers[3] has shown that a few studies had included control groups but very few among them had used matched controls to examine absconding events. Also, India has witnessed a significant shift in mental health care services over the last three decades. Earlier, only a few psychiatric hospitals provided mental health care services. With the establishment of general hospital psychiatry units at medical colleges, mental health care services have become more accessible than before. Mental health care services took a more deinstitutionalized approach, and all restrictive approaches were abolished from psychiatric care settings. Among a few studies from India, only one by Gowda et al.[24] recently examined the characteristics of the patient who abscond during inpatient care at a psychiatric hospital. Hence, the study was aimed to examine the sociodemographic and clinical profiles of adult inpatients who absconded in a specified period and to compare them with matched controls.

Methods

The study was conducted at a psychiatric hospital in India. Under adult psychiatry, the institute has three units for male patients, three units for female patients, and a dedicated unit for addiction psychiatry. Patients aged more than 18 years are admitted in these units. There is an open ward system, where the patients are admitted and managed with no/least restrictive measures. The institute has a boundary wall surrounding the campus and a main gate for entry and exit. The wards are adequately staffed with nurses and attendants across three fixed shifts of duty spanning over 24 hours. The staffing pattern is fixed for a type of shift (morning, evening, and night) across wards, based on the number of beds available for that ward. Each morning shift has 3–5 nursing personnel and 2–3 ward attendants, apart from the housekeeping staff, which is largely outsourced.

Approval was obtained from the Institutional Ethics Committee before collecting data. A separate register is maintained in the institute to record the absconding events. After an inpatient is reported missing in count from any ward during the mandatory headcount at the beginning and/or end of each shift, the information is passed to the treating team and hospital administration. A thorough search is conducted in the hospital premises. The local police station is also informed simultaneously. If the patient cannot be traced inside the hospital area and he/she fails to return to the hospital before midnight of the same day, either by himself/herself or by assistance, he/she is considered as absconding. Such events are recorded in the absconding register.

A list of case record file (CRF) numbers of patients who absconded between January 1, 2014 and December 31, 2018 was made from the register. Corresponding admission register of each case was accessed and searched for the admission period of that case. Parallelly, from other admitted inpatients of that admission period, a control was selected, matched with the diagnosis and age of the corresponding absconding case. Thus, each control was matched with the corresponding absconding case in the following order: (a) admission ward (admitted and residing in the same unit and ward as in each case of absconding) (b) admission period (the time period of admission during which the absconding case was admitted, (c) diagnosis, and (d) age. The method of chart review was followed for data collection. Sociodemographic information was recorded. Details for the event of absconding were collected. For clinical profiles, mental status examinations (MSE) done both during the admission and before the event of absconding were reviewed. As a part of the protocol of the institute, all the patients were examined regularly twice a day, in the morning and evening, and the clinical notes were recorded in CRFs. The MSE is structured and comprehensive in the case records. For cases, the most recent MSE before absconding was reviewed. For controls, the MSE most proximal to the date and time of absconding of the corresponding case was recorded. This was done to match immediate external factors around the event of absconding.

Statistical Analysis

Statistical analysis was performed, using the level of statistical significance of P < 0.05, in the Statistical Package for the Social Sciences (SPSS) version 22.0 for Windows. Sociodemographic characteristics and clinical profile of the sample were analyzed by descriptive statistics. Those records where either history was not available or MSE could not be elicited or did not specify psychopathology were excluded from the analysis. Chi-square test was used to assess discrete variables. Fisher’s exact test and Yate’s continuity correction was applied wherever applicable. After checking skewness, Student’s t or Mann–Whitney U test was applied for continuous variables.

Results

There was a total of 20,052 admissions in adult psychiatry in the specified period. During this period, 38 patients had absconded. Thus, the rate of absconding was 1.8/1000 admissions (0.18%).

Sociodemographic and Clinical Characteristics

Table 1 shows the sociodemographic characteristics of absconding cases and matched controls. The mean(±SD) age was 27.95 ± 6.96 years in the absconding cases. Most of the absconding cases were male (44.7%, n = 34). There was no difference in sex between the groups as the controls were selected from the same admission ward of the corresponding absconding cases. Other sociodemographic characteristics were comparable between the groups, except for a trend toward a more well-adjusted premorbid personality observed among controls (P = 0.05). Though patients with absconding had a higher proportion of substance use disorder as a comorbid diagnosis, no significant difference between the groups was observed (P = 0.49). Most cases in both groups had no physical comorbidity, and no significant difference was found between the groups (P = 1.00). The mean duration of stay in the hospital was 14.76 ± 12.31 and 41.68 ± 25.05 days in the absconding group till the event of absconding and the control group, respectively, with a significant difference between the groups (P < 0.001). Table 2 shows a comparison of the clinical profiles of the groups.

Table 1.

Comparison of Sociodemographic Variables between Groups (N = 76)

Characteristics		Absconding casesn = 38(%)	Controln = 38(%)	χ2/f/t/U	P
Type of admission	Voluntary	36(47.4)	38(50)	0.51y	0.47
	Under special circumstances	2(2.6)	0(0.0)
Age (in years)	Means ± SD	27.94 ± 6.96	28.02 ± 6.74	–0.05t	0.96
Age group(in years)	18–30	25(32.9)	26(34.2)	0.21f	1.00
	31–40	11(14.5)	10(13.2)
	≥41	2(2.6)	2(2.6)
Sex	Male	34(44.7)	34(44.7)	0.00y	1.00
	Female	4(5.3)	4(5.3)
Marital status	Unmarried	17(22.4)	18(23.7)	1.08f	0.81
	Married	21(27.6)	19(25)
	Others	0(0.0)	1(1.3)
Religion	Hindu	22(28.9)	29(38.2)	4.22f	0.13
	Muslim	15(19.7)	7(9.2)
	Others	1(1.3)	2(2.6)
Education	Illiterate	8(10.5)	2(2.6)	5.03f	0.09
	Below graduation	24(31.6)	32(42.1)
	Graduation and above	6(7.9)	4(5.3)
Occupation	Unemployed	13(17.1)	17(22.4)	1.19f	0.93
	Employed	2(2.6)	2(2.6)
	Labourer	9(11.8)	7(9.2)
	Students	4(5.3)	3(3.9)
	Others	10(13.2)	9(11.8)
Address	Rural	23(30.3)	26(34.2)	4.99f	0.07
	Semi urban	10(13.2)	3(3.9)
	Urban	5(6.6)	9(11.8)
Distance of address from hospital	<200 km	10(13.2)	9(11.8)	1.52	0.48
	201–400 km	12(15.8)	8(10.5)
	>400 km	16(21.1)	21(27.6)
State where the patient belongs to	Jharkhand	14(18.4)	12(15.8)	0.23	0.80
	Others	24(31.6)	26(34.2)
Family type	Nuclear	12(15.8)	9(11.8)	0.59	0.60
	Extended/joint	26(34.2)	29(38.2)
Socioeconomic status	Below poverty line	9(11.8)	5(6.6)	1.84f	0.42
	Above poverty line	27(35.5)	29(38.2)
	Not known	2(2.6)	4(5.3)
Family income (₹/month)	<10,000	26(34.2)	26(34.2)	2.32f	0.58
	10,000–50,000	8(10.5)	8(10.5)
	>50,000	2(2.6)	0(0.0)
	Not known	2(2.6)	4(5.3)
Informants	Parents	21(27.6)	20(26.3)	3.88f	0.29
	Siblings	7(9.2)	13(17.1)
	Spouse	4(5.3)	3(3.9)
	Relatives and others	6(7.9)	2(2.6)
Social support	Good/satisfactory	25(32.9)	26(34.2)	0.06	1.00
	Poor	13(17.1)	12(15.8)
Primary caregivers	Family	34(44.7)	37(48.7)	0.85y	0.35
	Relative and others	4(5.3)	1(1.3)
Family attitude	Concerned	23(30.3)	24(31.6)	0.05	1.00
	Ignorant	15(19.7)	14(18.4)
Relationship with family members	Cordial	13(17.1)	16(21.1)	0.50	0.63
	Strained	25(32.9)	22(28.9)

χ2: Chi-Square test, f: Fisher’s exact test, level of significance accepted at P < 0.05, y: Yate’s continuity correction. t: Student’s t test.

Table 2.

Comparisons of Clinical Profiles between Groups (N = 76)

Characteristics		Absconding casesn = 38(%)	Controln = 38(%)	χ2/f/t/U	P
Age of onset (years)	Means ± SD	20.92 ± 7.78	21.50 ± 6.53	–0.35t	0.72
Age of onset group	Before 18 years	15(19.7)	16(21.1)	0.05	1.00
	After 18 years	23(30.3)	22(28.9)
Duration of illness	<6 months	9(11.8)	4(5.3)	2.37f	0.37
	6 months to 2 years	4(5.3)	4(5.3)
	>2 years	25(32.9)	30(39.5)
Treatment history	No	13(17.1)	9(11.8)	1.02	0.44
	Yes	25(32.9)	29(38.2)
Past psychiatric illness	Absent	30(39.5)	27(35.5)	0.63	0.59
	Present	8(10.5)	11(14.5)
Past medical illness	Absent	33(43.4)	31(40.8)	0.39	0.75
	Present	5(6.6)	7(9.2)
Family history of psychiatric illness	Absent	19(25.0)	18(23.7)	0.00	1.00
	Present	19(25.0)	20(26.3)
Forensic history	Absent	33(43.4)	33(43.4)	0.00	1.00
	Present	5(6.6)	5(6.6)
Premorbid personalitya	Well adjusted	24(31.6)	31(40.8)	3.70y	0.05
	Not well adjusted	12(15.8)	4(5.3)
Number of hospitalizations	Single	25(32.9)	24(31.6)	0.05	0.81
	Multiple	13(17.1)	14(18.4)
Diagnosis	Affective disorders	14(18.4)	16(21.1)	1.81f	0.74
	Schizophrenia and related disorders	15(19.7)	15(19.7)
	Substance use disorders	7(9.2)	7(9.2)
	Others	2(2.6)	0(0.0)
Comorbid diagnosis	Affective disorders	4(5.3)	2(2.6)	3.44f	0.49
	Schizophrenia and related disorders	0(0.0)	1(1.3)
	Substance use disorders	15(19.7)	10(13.2)
	Others	4(5.3)	5(6.6)
	None	15(19.7)	20(26.3)
Comorbid physical illness	Present	9(11.8)	9(11.8)	0.00	1.00
	Absent	29(38.2)	29(38.2)
Treatment	Antidepressant/mood stabilizers only	0(0.0)	1(1.3)	1.17f	0.93
	Antipsychotics only	9(11.8)	10(13.2)
	ECT only	3(3.9)	3(3.9)
	Any combinations of above	26(34.2)	24(31.6)
Duration of stay in the hospital (days)	Means ± SD	14.76 ± 12.31	41.68 ± 25.05	198.5U	<0.001***
	Median	9.50	37.50
	Interquartile range	14.25	22.00

χ2: Chi-Square test, f: Fisher’s exact test, U: Mann–Whitney U, t: Student's t test, y: Yate's continuity correction. level of significance accepted at P < 0.05. ***P < 0.001. aFive patients were excluded from the analysis.

Mental Status Examination at the Time of Admission

More cases of absconding had loud and overproductive speech (32.9%, n = 25) than controls (25%, n = 19). There was a significant difference in affect between the groups at the time of admission (P = 0.005). More absconding cases (22.4%, n = 17) than controls (9.2%, n = 7) had irritable affect. More patients in the control group (n = 13) than in the absconding group (n = 5) had perceptual abnormalities (P = 0.04). Most of the cases in both the groups had impaired judgment and poor insight about illnesses at the time of admission. Table 3A shows the comparison of MSE findings between the groups at admission

Table 3A.

Comparison of Mental Status Examination at the Time of Admission between Groups (N = 76)

Characteristics		Absconding casesn = 38 (%)	Controln = 38 (%)	χ2/f	P
Motor behaviour	Increased	27(35.5)	22(28.9)	2.35f	0.36
	Decreased	1(1.3)	4(5.3)
	Normal	10(13.2)	12(15.8)
Speech	Loud and over productive	25(32.9)	19(25.0)	5.26f	0.09
	Mute/decreased productivity	2(2.6)	9(11.8)
	Normal	11(14.5)	10(13.2)
Affect	Euphoric/elated	12(15.8)	11(14.5)	13.74f	0.005**
	Irritable	17(22.4)	7(9.2)
	Depressed	0(0.0)	1(1.3)
	Dysphoric	6(7.9)	4(5.3)
	Constricted	3(3.9)	15(19.7)
Thought(stream, form and possession)	Poverty of thought/FTD	4(5.3)	8(10.5)	2.17f	0.62
	Thought alienation	3(3.9)	4(5.3)
	Obsessive and compulsive	1(1.3)	1(1.3)
	No abnormality detected	30(39.5)	25(32.9)
Thought (content) a	No abnormality detected	2(2.6)	0(0.0)	2.65f	0.26
	Delusions	27(35.5)	25(32.9)
	Depressive cognitions	3(3.9)	6(7.9)
Perceptual abnormalityb	Present	5(6.6)	13(17.1)	4.93	0.04*
	Absent	21(27.6)	14(18.4)
Somatic passivity	Absent	37(48.7)	36(47.4)	0.00y	1.00
	Present	1(1.3)	2(2.6)
Judgment	Impaired test, social, and personal	35(46.1)	34(44.7)	0.15f	1.00
	Impaired social and personal	3(3.9)	4(5.3)
	Impaired personal	0(0.0)	0(0.0)
	Intact	0(0.0)	0(0.0)
Insight	Grade-I	37(48.7)	33(43.4)	4.17f	0.11
	Grade-II	1(1.3)	4(5.3)
	Grade-III	0(0.0)	1(1.3)
	Grade-IV and above	0(0.0)	0(0.0)

χ2: Chi-Square test, f: Fisher’s exact test, level of significance accepted at P < 0.05, y: Yate’s continuity correction. *P<0.05, **P < 0.01. aThirteen patients were excluded from analysis. bTwenty-three patients were excluded from the analysis.

Mental Status Examination Before the Event of Absconding

There was a significant difference in affect between the groups (P = 0.001). Affect was found irritable among more cases of the absconding group (19.7%, n = 15) than the controls (3.9%, n = 3), while a higher proportion of constricted and poorly communicable affect was found in the controls (14.5%, n = 11) than the absconding group (3.9%, n = 3). A trend toward higher perceptual abnormalities was noted in the control group (P = 0.09). There was a significant difference between the groups in relation to judgment and insight (P < 0.001). Impaired test, social, and personal judgments were found more in the absconding group (32.9%, n = 25) than the controls (11.8%, n = 9). Most of the patients in the absconding group had no insight (47.4%, n = 36). Table 3B shows the comparison of MSE findings between the groups before the time of absconding.

Table 3B.

Comparison of the Latest Mental Status Examination Before Leave (N = 76)

Characteristics		Absconding casesn = 38 (%)	Controln = 38 (%)	χ2/f	P
Motor behaviour	Increased	10(13.2)	12(15.8)	0.36f	0.93
	Decreased	3(3.9)	3(3.9)
	Normal	25(32.9)	23(30.3)
Speech	Loud and over productive	12(15.8)	14(18.4)	2.14f	0.58
	Mute/decreased productivity	4(5.3)	7(9.2)
	Normal	22(28.9)	17(22.4)
Affect	Euphoric/elated	12(15.8)	10(13.2)	18.83f	0.001**
	Irritable	15(19.7)	3(3.9)
	Depressed	1(1.3)	3(3.9)
	Dysphoric	7(9.2)	6(7.9)
	Constricted	3(3.9)	11(14.5)
	Euthymic	0(0.0)	5(6.6)
Thought(stream, form and possession)	Poverty of thought/FTD	3(3.9)	2(2.6)	0.91f	1.00
	Thought alienation	1(1.3)	2(2.6)
	Obsessive and compulsive	1(1.3)	1(1.3)
	No abnormality detected	33(43.4)	33(43.4)
Thought (content)a	No abnormality detected	7(9.2)	12(15.8)	2.23f	0.31
	Delusions	25(32.9)	19(25.0)
	Depressive cognitions	4(5.3)	3(3.9)
Perceptual abnormalityb	Present	5(6.6)	11(14.5)	3.16	0.09
	Absent	30(39.5)	23(30.3)
Judgment	Impaired test, social, and personal	25(32.9)	9(11.8)	17.47f	<0.001***
	Impaired social and personal	11(14.5)	15(19.7)
	Impaired personal	2(2.6)	13(17.1)
	Intact	0(0.0)	1(1.3)
Insight	Grade-I	36(47.4)	18(23.7)	21.64f	<0.001***
	Grade-II	1(1.3)	14(18.4)
	Grade-III	1(1.3)	5(6.6)
	Grade-IV and above	0(0.0)	1(1.3)

χ2: chi-square test, f: Fisher’s exact test, level of significance accepted at P < 0.05, FTD: formal thought disorder. **P < 0.01, ***P < 0.001. aSix patients were excluded from analysis. bSeven patients were excluded from the analysis.

Factors Related to Events of Absconding

Table 4 shows various factors related to the absconding behavior. The mean duration of stay till the date of absconding was 14.76 ± 12.31 days. The most common timing was 8 am–2 pm (28.95%, n = 11). The most common mean chosen was scaling the boundary wall (34.21%, n = 13), followed by escaping through the main gate (21.05%, n = 8). For 28.94% (n = 11) of the absconding cases, the means chosen for absconding could not be recorded as the patients, when traced later, did not share the details of the same.

Table 4.

Factors Related to Absconding (N = 38)

Characteristics		Number(%)
Time of missing	3 am—8 am	10(26.32)
	8 am—2 pm	11(28.95)
	2 pm—8 pm	9(23.68)
	8pm–3am	8(21.05)
Mean chosen to abscond	Through the main gate	8(21.05)
	Scaling the boundary wall	13(34.21)
	By manipulating workers	1(2.63)
	During referral to a medical centre for management of a physical illness	2(5.26)
	By manipulating other patients/their guardians	3(7.89)
	Details unspecified	11(28.94)
Past history ofabsconding	No	34(89.47)
	Yes	4(10.53)
Duration of stay before absconding (days)	Means ± SD	14.76 ± 12.31
	Median	9.50
	Interquartile range	14.25

Discussion

Our study examined all the absconding events from the inpatient care of a psychiatric hospital over a period of five years. The strengths of our study were not missing any cases of absconding and close matching between the cases and controls. Admission period and admission ward were matched, as these might influence immediate environmental factors such as, level of supervision, security level, and clinical management protocols. As psychiatric diagnoses are often heterogeneous in nature and dependent upon a constellation of psychopathologies, matching diagnosis helped us examine several clinical factors, including psychopathologies not attributable to the difference in the diagnosis between the groups. A similar matching protocol had been adopted by research that investigated absconding from an inpatient setting and other uncommon hospital events like inpatient suicide.[8, 15, 25–29] As the controls were matched for the admission ward and admitted under the same unit of the corresponding absconding case, both the groups became matched for sex too. Another strength was that we reviewed the MSE at two time points.

The rate of absconding in our study was 1.8 incidents per 1,000 admissions. This rate is low compared to previous studies, where the rate ranged from 1.85% to 17.2%.[20, 22, 23, 30, 31] A systematic review found the rates to be substantially lower for locked wards (1.34/100 admissions) compared to open wards (7.96/100 admissions). Despite being an open ward system, comprehensive management, including continuous risk assessment, monitoring, and supervision by treating teams consisting of a psychiatrist, clinical psychologist, psychiatric social worker, psychiatric nurse, and ward attendants might have reduced absconding, accounting for the lower rate in our study. Different rates across the world might be due to a difference in the type of security measures, type of hospital care, the presence of forensic patients, legal measures, and multiple other factors.[3]

Most of the total sample in our study was admitted as voluntary admission under Mental Health Act 1987 and subsequently as independent admission under Mental Healthcare Act 2017 in both the cases and the control group. In contrast to this, most studies have demonstrated an increased prevalence of involuntary admission among inpatients who abscond.[4, 7, 13, 32] A recent study by Gowda et al.,[24] at a psychiatric hospital, also showed that most patients who absconded had involuntary admission. However, a case-control study by Bowers et al.[5] did not demonstrate the above finding and instead suggested that the escaping incidents might be more officially reported among involuntary patients than the voluntary ones. In our study, the mean hospital stay for the absconding group was 14.8 days. Though a recent study found that a longer duration of stay is predictive of absconding,[31] other studies found that most absconding occurred during the first two weeks following admission. This may be because, during the initial two weeks of inpatient care, adjusting to the hospital environment happens and the illness is often most acute.[19, 33]

We did not find any significant difference regarding marital status between the groups. A recent case-control study by Khammarnia et al.[15] showed no significant difference between those who absconded and those who did not, in terms of age and marital status. A few studies have found a higher rate of absconding among unmarried inpatients.[4, 33] Our study did not find any difference in religion between the groups, which is consistent with the findings from other studies.[4, 21] However, some studies have a different finding.[5, 34–36] Like other studies, most of the absconding cases in our study were male and from a younger age group.[15, 22, 24, 30, 33] Similarly, John et al.[30] also found a smaller proportion of females than males and speculated that the low rate among the women might be due to the culturally inculcated restrictions among Indian women. On the contrary, Khisty et al.[19] and Dickens and Campbell[4] did not find any significant difference between absconding cases in men and women. Cultural shift over time or difference in the setting might have caused this finding.[19]

While Gowda et al.[24] found that most patients who absconded belonged to lower socioeconomic status, our study did not find any such difference between the groups. In our study, most of the patients in the absconding group belonged to joint or extended family. In contrast, Gowda et al.[24] had found that most of them belonged to a nuclear family.

The absconding patients and the controls did not differ in the age of onset, duration of illness, history of past treatment, history of psychiatric or medical illness, history of family illnesses, or the presence of forensic history.

The most common diagnosis in the absconding group was schizophrenia and related disorders, followed by affective disorders and substance use disorders. This finding is also supported by previous descriptive and case-control studies, which found psychotic illness to be a common factor predicting absconding events.[5, 12, 22, 24, 31, 37] A few studies have also found increased rates of affective illness, substance use, and personality disorders.[10, 38, 39]

A significantly higher proportion of patients who absconded had irritable affect both at the time of admission and around the time of absconding. Also, those who absconded had less perceptual abnormalities than controls at admission, which could explain the absence of perceptual abnormality among the absconding cases at the time of absconding too. This is consistent with Khisty et al.[19] who speculated that being more guarded than controls might be the reason why they were less likely to report hallucinations.

There was a significant difference in judgment and insight around the event of absconding between the groups. Patients of the absconding group had grade I insight around the time of absconding, while the patients of the control group had shown improvement to have grade II or better insight (P < 0.001). A recent study from a psychiatric hospital in India also found a higher prevalence of absconding among patients who had an absent insight.[24]

About 10% of cases had a history of prior absconding from the hospital in the previous admissions. This proportion is less than what was found by Meehan et al.,[40] who found that one-third of those who absconded had a similar history in their previous hospitalizations. The reason behind this discrepancy might be the improvement in the psychiatric management of inpatients, including the physical structure and environment of the psychiatric care facility.

Most of the events occurred during the daytime and early morning or evening hours, as compared to early morning clustering observed in another study.[40] Khisty et al.[19] found that most incidents occurred during the daytime (09:00 hr to 18:00 hr) Several factors might contribute to this. Leaving the hospital ward may be easier during the daytime because at that time, almost in any hospital, much more people move around or visit, thereby making it less difficult to make way for the exit. Dickens and Campbell[4] reasoned that often, at nighttime, even open wards are locked by night staff, which seems to be relevant in our study too. A literature review[3] did not find any common patterns, which might be due to the variations in the hospitals, organization of shifts among hospitals, settings, hospital protocols, and countries.

The most common mean chosen for absconding was scaling through the boundary wall, followed by escaping through the main gate, manipulating other patients or their guardians, during a referral to other medical centers for physical illness, and by manipulating the workers. A study by Yasini et al.[33] found that high numbers of absconding take place from the main door of the hospital, followed by other sites (walls, windows, etc.). Agreeing with previous researchers, we speculate that increased staffing would help by improving the monitoring and observation of inpatients and the understanding of their psychological and nursing care needs.[33, 41]

Our study suffered from a few limitations. For about one-third of the absconding cases, the means chosen to abscond were not shared by the patients and hence could not be recorded. Reasons for absconding have not been examined as well in our study. Qualitative information could not be examined. We included a modest period of observation of five years only. Limitations of a retrospective design based on CRF review need a mention.

Conclusion

Most of the inpatients who absconded were male, from a younger age group, from joint or extended family, diagnosed with schizophrenia or mood disorder, and having comorbid substance use disorder, irritable affect, impaired judgment, and absent insight. Most incidents of absconding happened during the first two weeks following admission. About one-tenth of those who absconded had a similar history. Most of the events occurred during daytime and early morning or evening hours, and the most common means chosen were scaling the boundary wall and passing through the main gate. Our study adds to the existing sparse literature on the nature and pattern of absconding and the profile of those inpatients who abscond from a psychiatric hospital. Further research should focus on this event with a longer period of observation and should include qualitative design. This would lead to a more effective risk assessment in psychiatric inpatient care and facilitate reintegration into the community.