- Email: [email protected]
Polydipsia and schizophrenia in a psychiatric hospital: a replication study
Schizophrenia Research 57 (2002) 293 – 301 www.elsevier.com/locate/schres
Polydipsia and schizophrenia in a psychiatric hospital: a replication study Jose de Leon a,*, Joseph Tracy b, Eileen McCann a, Amy McGrory a a
Mental Health Research Center at Eastern State Hospital, 627 West Fourth St., Lexington, KY 40508, USA b Thomas Jefferson University, Jefferson Medical College, Philadelphia, PA, USA Received 20 February 2001; accepted 4 June 2001
Abstract A prior study in a US state hospital suggested that schizophrenia, smoking and long hospitalization were associated with polydipsia. This study, in another US hospital, attempts to (1) replicate that schizophrenia and smoking are associated with polydipsia, and (2) rule out that this relationship is partly explained by alcohol and drug use. Both studies have similar methodologies. The second sample included 588 inpatients. Models of variables associated with polydipsia were developed using logistic regression. In the second study, after correcting for other factors, the association between polydipsia and schizophrenia showed a borderline significance, while polydipsia and smoking displayed a significant association. Neither organic brain lesions, nor alcohol or drug use, were associated with polydipsia. An analysis combining both samples showed that: (1) schizophrenia, long hospitalization, smoking and heavy smoking were significantly associated with polydipsia, and (2) male gender and Caucasian race (but not smoking) increased the risk of developing water intoxication in polydipsic patients. These two studies in severely mentally ill patients suggest that the association of polydipsia with schizophrenia, smoking and chronicity is consistent and independent from the definition of polydipsia (by staff, a biological method or the combination of both). Psychiatric medications do not appear to explain most cases of polydipsia in these patients. D 2002 Elsevier Science B.V. All rights reserved. Keywords: Schizophrenia; Polydipsia; Water intoxication; Nicotine; Smoking; Logistic regression
1. Introduction Polydipsia, among chronic psychiatric patients, is poorly understood and underdiagnosed. Patients may have simple polydipsia, with accompanying polyuria, polydipsia, with water intoxication, and/or physical
*
Corresponding author. Tel.: +1-859-246-7487; fax: +1-859246-7019. E-mail address: [email protected] (J. de Leon).
complications secondary to ingestion of fluids in large quantities (de Leon et al., 1994a). Ingestion of lithium or diuretics, and endocrine disorders such as diabetes mellitus, are causes of polyuria, frequently found among chronic psychiatric inpatients. In all of these polyuric conditions, the primary abnormal mechanism is polyuria with a compensatory or ‘‘secondary’’ polydipsia. (For simplification, in this article, those types of etiologies are designated as medical causes of polyuria). On the other hand, the typical polydipsia that induces a
0920-9964/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 0 - 9 9 6 4 ( 0 1 ) 0 0 2 9 2 - 4
294
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
secondary polyuria (seen in psychiatric patients) will be termed ‘‘primary’’ polydipsia. In almost all cases of polydipsia with water intoxication, there is an increased release of antidiuretic hormone (ADH) (Goldman et al., 1997). Alternatively, psychiatric patients without polydipsia may develop a syndrome of inappropriate secretion of ADH (SIADH) induced by drugs such as carbamazepine or sulfonylureas. It is also possible that SIADH may occur in non-polydipsic patients in the absence of unrecognized causes. Hyponatremia in psychiatric patients may also be precipitated by other factors which may decrease free-water excretion, such as thiazide diuretics or hypothyroidism (Goldman, 1991). 1.1. Polydipsia According to a review of the literature, primary polydipsia may be present in more than 20% of chronic inpatients (de Leon et al., 1994a). In a recent French study, the staff reports of 253 chronic patients were reviewed. The study results showed a 13% rate of polydipsia among the sample and a 10% rate of primary polydipsia (Mercier-Guidez and Loaz, 2000). Little is known about what factors contribute to polydipsia. Chronicity of illness (Evenson et al., 1987; de Leon et al., 1996), and surprisingly, smoking (de Leon et al., 1996) may be associated with polydipsia. The literature has suggested that heavy smoking is only associated with the development of water retention and water intoxication in polydipsic patients but a recent French study (Mercier-Guidez and Loaz, 2000) confirmed our findings that there is an association between polydipsia and smoking. 1.2. Polydipsia with water intoxication Previous surveys suggest that an average of 29% (25 – 86%) of patients with polydipsia have a history of water intoxication (de Leon et al., 1994a). However, in our first survey, which diagnosed a larger number of polydipsic patients than in prior studies, the prevalence of a history of water intoxication was only 14% (de Leon et al., 1996). A more recent survey suggested a prevalence of 31% (Mercier-Guidez and Loaz, 2000). Drugs that decrease free-water excretion (thiazide diuretics, sulfonylureas, and carbamazepine) may
contribute to the development of water intoxication in patients with polydipsia. Additionally, Caucasian race, male gender, chronicity, a diagnosis of schizophrenia and smoking may also be associated with the development of water intoxication in patients with polydipsia (Evenson et al., 1987; Shah and Greenberg, 1992; de Leon et al., 1994a, 1996). These factors may influence each other but it is not clear which of these factors may have a primary influence. Our first epidemiological study, which included all inpatients at a psychiatric hospital in Philadelphia, was important in order to establish that polydipsia is associated with schizophrenia and smoking after correcting for the effect of other factors, such as chronicity. No replications of this study have been published using multivariate analyses to correct for the effect of other factors. Moreover, our study did not correct for the effects of abuse of other substances. Some data in the literature suggests that schizophrenic patients with substance abuse disorders demonstrate high rates of polydipsia and water intoxication (Ripley et al., 1989). It has been found that patients with schizophrenia tend to abuse alcohol and drugs more frequently than the normal population (Buckley, 1998). This current study, at a second US psychiatric state hospital near Philadelphia, tried to (1) replicate the data showing that schizophrenia and smoking are associated with polydipsia; and (2) rule out that this relationship is partly explained by abuse of alcohol and/or drugs.
2. Methods The methodology applied to both studies was very similar (de Leon et al., 1996). The first study was performed in 1990 and included all patients admitted at Haverford State Hospital located in suburban Philadelphia. This second replication study was performed in 1992 at Norristown State Hospital, also in suburban Philadelphia. Both hospitals serviced Philadelphia and the surrounding counties but their catchment areas did not overlap. 2.1. Subjects Out of the 659 patients in the hospital, 588 agreed to participate in this study. Of those 588 patients, a few either did not provide all of the requested infor-
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
mation, or cooperated with only some of the procedures. As expected, this affected the sample size in some of our analyses. The sample in this study has been previously described (de Leon et al., in press). Sixty-three percent (370/588) of the patients were male. Seventy-one percent (415/588) of the patients were Caucasian, 27% (161/588) were African-American and 2% (12/ 588) were from other races. The mean patient age was 46 years (standard deviation [S.D.] 16.1). Twentyeight percent (166/588) of the patients were older than 55 years of age. The mean duration of current hospitalization was 9.2 years (S.D. 15.0). Most patients (90%) were taking antipsychotics. Mean dose in chlorpromazine equivalents was 931 (S.D. 845). The percentage of patients with history of alcohol abuse or dependence was 39%. The percentage of patients with history of abuse or dependence on any drug was 31%. Seventy percent of the patients were smokers, and of those, 17% were heavy smokers ( z 1.5 packs/day). Clinical diagnoses were taken from the charts. The DSM-IIIR diagnoses were as follows: schizophrenia (71%), schizoaffective disorder (7%), organic mental disorders (8%), affective disorders (5%) and other disorders (9%). As in the first study, the diagnoses were divided into schizophrenic (schizophrenic or schizoaffective) and non-schizophrenic categories, 78% and 22%, respectively. Antipsychotic medications were converted into chlorpromazine equivalents, with dosages higher than 1000 mg of chlorpromazine equivalents classified as ‘‘high’’. A table was used to convert antimuscarinic activity of psychotropic medication to benztropine equivalents (de Leon et al., 1994b). Dosages equal to or greater than 2 mg of benztropine equivalents were classified as ‘‘moderate’’. 2.2. Assessment of polydipsia The AM weight was obtained between 7 and 8 AM, after the patient voided and before breakfast. After ensuring that patients were in the same state of dress, the PM weight was obtained between 3 and 4 PM. Normalized diurnal weight gain (NDWG), which is the percentage of weight gain from morning to afternoon, was derived by dividing the afternoon weight gain by the morning weight and multiplying by 100 (Vieweg et al., 1989).
295
The NDWG is considered a measure of polydipsia and water retention. In normal individuals, NDWG is lower than 1.2%. An NDWG > 4% is associated with a decrease in serum sodium by at least 10 mM/l and carries a serious risk of water intoxication (Vieweg et al., 1989). AM and PM urine samples were obtained, and the SPGU was measured with a refractometer. SPGU lower than 1.009 is suggestive of polydipsia (Blum et al., 1983). Staff (nurses and nurses’ aides) were also interviewed regarding patients who were drinking excessive amounts of fluid (Luchins et al., 1992; Tracy et al., 1996). 2.3. Definitions of polydipsia and water intoxication A diagnosis of polydipsia by staff or an SPGU lower than 1.009 was considered a positive case of polydipsia (de Leon et al., 1996). Some polydipsic patients may have secondary polydipsia explained by polyuria induced by lithium, diuretics or diabetes mellitus. The presence of any of these causes of polyuria was considered as a confounding variable in the logistic regression models. In the prior study, a history of water intoxication was established by an extensive chart reviewed by a physician, however, this was not an option in this study. To establish which factors were associated with a high risk of hyponatremia among patients with primary polydipsia, those patients were subdivided using the NDWG. Polydipsic patients with very high NDWG (>4.0%) were classified as displaying a very high risk of developing water intoxication. 2.4. Statistics The Statistical Package for Social Sciences (SPSS) was used for calculations. Two-way cross-tabulations for univariate analyses (Table 2) computed the odds ratios (ORs). Univariate analyses are described to compare with other scientific reports that may not use logistic regression. The 95% confidence intervals (CI) for ORs were calculated. Logistic regression analyses of the probability that a patient was polydipsic were corrected by using several independent variables: diagnosis (schizophrenic or non-schizophrenic group), smoking, heavy smoking ( z 1.5 packs/day or not), high dose of antipsychotic medi-
296
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
hospitalization (4.8 years), and dose in chlorpromazine equivalent (732) were also lower. The samples of both hospitals were combined in a logistic regression analysis and the effect of these clinical variables was controlled. The differences between the two hospitals should not affect comparisons between the first and replication studies; the logistic model allows for studying the effect of a particular variable while correcting for the effects of other variables. Logistic regressions that included history of alcohol or drug abuse disorders and CNS abnormalities were performed only in this replication study.
cation (>1000 chlorpromazine equivalents or not), moderate doses of antimuscarinic drugs ( z 2 benztropine equivalents or not), old age (>55 years or not), gender, race, long duration of hospitalization (>2 years or not), and causes of polyuria. Any patient with history of substance abuse or dependence in their chart was considered positive for alcohol or drug disorders. By chart review, history of any ‘‘organic’’ central nervous system (CNS) abnormality was also established in 36% of patients. The Hosmer –Lemeshow Goodness-of-Fit test was used to test the fitness of the logistic models; all fit well. Logistic regression analyses for polydipsia are reported in Table 3. Similar analyses were performed within the polydipsic patient sample. After excluding medical causes of water retention (hypothyroidism or treatment with carbamazepine, sulphonylureas or thiazides), polydipsic patients were divided into those with a high risk of developing water intoxication (NDWG>4.0%) and those without. There were some differences between the first study and this study: the percentage of Caucasians in the first hospital was higher (84%), but the percentage of elderly patients (15%) and patients with schizophrenia (65%) were lower. Additionally, in the first hospital the mean age (42 years), duration of
3. Results 3.1. Polydipsia 3.1.1. Frequency of polydipsia The overall frequency of abnormalities in SPGU was a little higher in this second study in both the total and the schizophrenic sample (Table 1). Thus, the percentage of people with polydipsia, when determined by combining staff report and SPGU, was slightly higher in the second hospital (49%, CI 45–
Table 1 Frequencies of polydipsia and high risk of water intoxication Polydipsia All sample
Staff diagnosis AM SPGU < 1.009 PM SPGU < 1.009 SPGU < 1.009a Combining staff and SPGU No medical causes Medical causes
Schizophrenia
First
Replication study
First
Replication study
26% 20% 29% 34% 42% 19% 23%
24% 33% 43% 47% 49% 34% 15%
33% 33% 35% 39% 48% 23% 25%
28% 36% 46% 50% 53% 37% 16%
(142/588) (139/428) (150/346) (221/466) (290/588) (200/588) (90/588)
(124/449) (117/330) (123/269) (179/359) (237/449) (166/449) (171/449)
Polydipsia with high risk of water intoxication All sample
Very high NDWG (>4.0%) With causes of water retention Without causes of water retention a
Lower of AM and PM SPGUs.
Schizophrenia
First
Replication study
First
Replication study
5.6% 0.3% 5.3%
4.4% (26/588) 0.7% ( 4/588) 3.7% (22/588)
8.0% 0.4% 7.6%
4.7% (21/449) 0.7% (3/449) 4.0% (18/449)
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
53% versus 42%, CI 37– 47%). However, this difference was not significant. 3.1.2. Univariate analyses of polydipsia The results from the univariate analyses were remarkably similar to those of the first study in which schizophrenia, smoking, heavy smoking, and causes of polyuria were also significantly associated with polydipsia. Other variables were not so consistent in both studies (Table 2). 3.1.3. Logistic regression for polydipsia Remarkably, there were many similarities in the variables predicting polydipsia in both studies’ total samples. For instance, causes of polyuria were significant in both studies, as expected; schizophrenia had borderline significance after correcting for other factors; and heavy smoking was significant in the first study but only borderline in the second. Finally, when both studies were combined and the effects of the hospital and the causes of polyuria were corrected, schizophrenia, long hospitalization, smoking and heavy smoking were all significant factors in predicting diagnosis of polydipsia (Table 3). In this second study, we found that smoking was associated with polydipsia in the schizophrenic sample. Additionally, long hospitalization, smoking and heavy smoking were all significant factors after correcting for other variables and combining both hospi-
297
tal samples. In the non-schizophrenic sample, long hospitalization and causes of polyuria were significant in both hospitals. After combining both hospitals, heavy smoking also became significant. 3.1.4. Correcting for other factors In the second study, history of alcohol and drug abuse disorders was never a significant factor in univariate analyses or logistic regression with polydipsia as the dependent variable. The presence of any kind of ‘‘organic’’ CNS abnormality was also non-significant. 3.2. Polydipsia with high risk of developing water intoxication 3.2.1. Frequency of polydipsia with high risk of developing water intoxication There was a slightly lower frequency of polydipsia with water retention in the second study. Note there is no way of estimating how uncooperativeness may have influenced these differences (Table 1). 3.2.2. Univariate analyses of polydipsia with high risk of developing water intoxication In the first study, male gender, a diagnosis of schizophrenia, high doses of antimuscarinic drugs, race and prolonged hospitalizations were associated with high NDWG in patients with polydipsia, after excluding causes of water retention. In this replication
Table 2 Odds ratios (and confidence intervals) of significant variables associated with polydipsia and with high risk of water intoxication in univariate analyses High risk of water intoxication in polydipsic patients
Polydipsia
Schizophrenia Smoking Heavy smoking High antipsychotic dose Antimuscarinic drugs: moderate dose Male gender Old age Caucasian race Long hospitalization Causes of polyuria
First study
Replication study
2.2 2.4 4.3 1.8 1.6
1.8 (1.2 – 2.6) 2.9 (2.0 – 4.2) 2.6 (1.5 – 4.6)
0.54 b.s.
1.5 (1.1 – 2.1) 0.72 (0.50 – 1.0) b.s.
1.9 2.9
1.5 (1.0 – 2.1)
First study
Replication study
4.0 2.8 b.s.
* 4.0 (1.1 – 14.0)
4.3
*
Only significant results are reported in the table. Blank spaces indicate non-significant odds ratios. b.s.: borderline significance. * Not significant but results were in the opposite direction with p-values 0.05 – 0.10.
298
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
Table 3 Odd ratios (and confidence intervals) of significant variables associated with polydipsia in logistic regression models Total sample First study
Schizophrenia Smoking Heavy smoking Old age Long hospitalization Causes of polyuria
Replication study
Combination of both studiesa
N = 360
N = 536
N = 896
1.4 b.s.
1.5 (0.97 – 2.3) b.s. 2.3 (1.5 – 3.4) 1.7 (0.97 – 3.1) b.s.
1.5 (1.1 – 2.1) 2.0 (1.4 – 2.8) 2.5 (1.7 – 3.8)
1.6 (1.1 – 2.3)
1.5 (1.1 – 2.1) 1.9 (1.4 – 2.6)
Replication study
Combination of both studiesa
3.4 0.29 1.9 2.6 Schizophrenic sample First study N = 237
Smoking Heavy smoking Old age Antimuscarinic drugs: Moderate dose Long hospitalization Causes of polyuria
N = 416
N = 652
2.5 (1.5 – 4.0)
2.1 (1.4 – 3.2) 2.3 (1.5 – 3.5)
3.3 0.42 b.s. 1.9 2.1 2.1
1.5 (1.1 – 2.1) 1.5 (1.1 – 2.2)
Non-schizophrenic sample
Smoking Heavy smoking Long hospitalization Causes of polyuria
First study
Replication study
Combination of both studiesa
N = 123
N = 120
N = 243
4.3 2.2 b.s. 5.4
2.2 (0.90 – 5.4) b.s. 3.4 (1.4 – 8.4)
3.8 (1.5 – 9.4) 1.9 (1.0 – 3.5) 4.0 (2.1 – 7.5)
Only significant results are reported in the table. Blank spaces indicate non-significant odds ratios. b.s.: borderline significance. a The second hospital had association with polydipsia: corrected OR was 1.9 (1.4 – 2.6) in the total sample, 1.8 (1.2 – 2.5) in the schizophrenic sample and 2.3 (1.2 – 4.2) in the non-schizophrenic sample.
study, only male gender was associated with high NDWG. Long hospitalizations and moderate doses of antimuscarinic drugs had a borderline significance. However, these two latter factors were associated with low NDWG (Table 2). 3.2.3. Logistic regression for polydipsia with high risk of developing water intoxication Logistic regressions of the first and second hospital showed very similar results to univariate analyses (Table 2). When both samples were combined (allowing greater statistical power) and all variables included in Table 2 were tested, only male gender was significant (OR = 2.6, CI 1.1 –6.3). Caucasian race was close to significance (OR = 3.0, CI 0.97 – 9.3).
4. Discussion 4.1. Limitations This study has several limitations, including a small amount of missing data, the lack of repeated biological measures, and the lack of a thorough chart review of all available records to provide figures of clearly identified episodes of water intoxication (5% in the first study). Missing data in some variables may influence estimations of prevalence, but it is important to remember that this study included a large sample. The staff at both hospitals had known these patients in these studies for many years. We included staff reports in the diagnosis of polydipsia because we believe these
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
practice results in a substantial decrease in false negatives from patients who were not drinking too much the day of the survey. Of course, it is possible that this may led to some false positives, however, based on our observations in both hospitals, the staff tended to report false negatives and they did not identify some of the obvious polydipsic patients who may have been drinking when off the units. However, staff diagnoses appear to have a lower number of false positives of polydipsia in general (staff members do not identify non-polydipsic patients as polydipsics). On the other hand, the staff may not be able to distinguish primary and secondary polydipsia. To ensure that the definition of polydipsia was not an issue, we repeated the analyses after combining both studies using either only staff diagnosis or biological diagnosis. Using only the staff diagnosis, the OR was 1.7 for schizophrenia, 1.8 for smoking, 2.7 for heavy smoking and 2.2 for long hospitalization. Using only the SPGU diagnosis, the OR was 1.5 for schizophrenia, 1.6 for smoking, 2.4 for heavy smoking and 1.4 for long hospitalization. These results are similar to those described in Table 3 using the combined definition of polydipsia (staff and SPGU). It is even more remarkable that two new analyses, using two complete independent and unrelated methods to diagnose polydipsia, staff diagnose and SPGU, came up with the same common variables (schizophrenia, smoking, heavy smoking and long hospitalization) associated with polydipsia. In summary, these two surveys in two US hospitals should be considered unique and valuable data because it is increasingly difficult to amass such surveys since fewer and fewer psychiatric patients spend many years within the confines of a psychiatric hospital. Also, the current ethical research regulations make these noninvasive naturalistic studies more difficult. 4.2. Strengths However, in spite of its limitations, one must remember that our first study is the only epidemiological study of a large and complete sample of severely mentally ill patients, using a multivariate method (logistic regression) to explore which variables are associated with polydipsia. Replication studies are frequently used in scientific methodology to test for the robustness of the scientific findings. Results that tend to be replicated in several
299
studies probably reflect ‘‘true’’ associations. Results that are not replicated in later studies or with different methods may reflect spurious or chance findings. In summary, we believe that the replication, by use of the similar methodology in both studies, helps to strengthen the reliability of the results. Moreover, the combination of both samples provided greater statistical power. Schizophrenia, smoking and chronicity appear to be consistently associated with polydipsia across the tables. This was also true when polydipsia was only defined by the staff or only by SPGU. Psychiatric medications do not appear to explain most cases of polydipsia. However, lithium and medical causes of polyuria appear to be more powerful predictors of polydipsia in non-schizophrenic patients. 4.3. Differences between the two hospitals After correcting for the effect of confounding variables, the odds of polydipsia in the second hospital was 1.8 times higher than in the first (Table 3). The effect of larger chronicity of the patients in the second hospital may not explain this difference since the logistic regression model will correct for it. It cannot be ruled out that a better measure of chronicity (such the unavailable duration of unremitting illness or active psychosis) may explain this difference between these two studies. It is possible that the refractometers for measuring the SPGU had different sensitivities, but this was not tested. Regardless, the SPGU is a relatively ‘‘gross’’ method. Better methods, such as urine osmolality concentration of creatinine, were not available. 4.4. Polydipsia and schizophrenia It is not known why schizophrenia may be associated with polydipsia. The temporal lobe abnormalities that are particularly found in schizophrenic patients have been posited to be important in patients with polydipsia and hyponatremia (Luchins et al., 1997). However, this will not explain simple polydipsia. 4.5. Polydipsia and smoking The association between polydipsia, smoking, and heavy smoking is remarkable. Particularly, it is striking that both have additive factors. Since nicotine increases ADH secretion, it is expected that smoking
300
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
may contribute to water retention and hyponatremia in polydipsic patients (de Leon et al., 1994a). However, these two studies suggest the association appears to be between smoking and increased fluid intake. The physiological literature does not provide clues as to why smoking may be associated with an increased fluid intake. However, Adler et al. (1998) described that the subfornical organ involved in the regulation of water drinking is rich in nicotine receptors. Literature coming from experimental psychology field has led to suggest that both smoking and polydipsia may be repetitive behaviors associated to situations of low reinforcement. The association between polydipsia and other repetitive behavior has been tested in two types of designs: within and across individuals. Using a within-individual design, Shutty et al. (1995) found that in two polydipsic patients there was a temporal relationship with stereotyped grooming and pacing. Using an acrossindividual design in the schizophrenic patients of our second hospital, we found that other repetitive behavior was not related to polydipsia defined by staff but smoking was indeed associated to polydipsia (Tracy et al., 1996). In a carefully controlled study of 10 polydipsic patients with history of hyponatremia, Shutty (1996) found that cigarette use was correlated with the total amount drunk. So in summary, this behavioral literature suggest that smoking and water drinking may be repetitive behaviors that are in some way associated but it is not clear whether polydipsia is associated with other repetitive behaviors or not. 4.6. Polydipsia and psychiatric medications Before any psychiatric medications were available, Sleeper (1935) demonstrated by catheterization procedures that 25% of 92 male schizophrenics were polydipsic. This classical study proves unequivocally that primary polydipsia in schizophrenia cannot be explained by antipsychotic medications. Certainly, polydipsia explained by lithium (or diuretics) may explain some cases of polydipsia, particularly, among non-schizophrenic patients. The presence of lithium treatment does not completely rule out primary polydipsia; moreover, lithium has been used to decrease the risk of hyponatremia in primary polydipsic patients (Vieweg et al., 1988).
4.7. Polydipsia and alcohol and drug abuse Alcohol and drug use did not appear to be associated with polydipsia in the second hospital. Replication from an independent study may be needed. 4.8. High risk of developing water intoxication in polydipsic patients Obviously, polydipsic patients taking any medication that is either associated with SIADH (carbamazepine or sulphonylureas) or decreases free-water clearance (thiazides) is at increased risk of developing water intoxication. In our first study, history of water intoxication by chart review appeared to be associated with longer hospitalization (de Leon et al., 1994a). The prior literature suggested that male gender and Caucasian race (versus African-American race) might also be associated with water intoxication. In our first study, both appear to be overrepresented but did not reach significance in the small sample studied. Both male gender and Caucasian race appeared to be associated with high risk of developing water intoxication when both sets of samples were combined.
5. Conclusion In summary, these two studies in severely mentally ill patients suggest that the association of polydipsia with schizophrenia, smoking and chronicity is consistent and independent from the definition of polydipsia (by staff, a biological method or the combination of both). Psychiatric medications do not appear to explain most cases of polydipsia in these patients. Organic brain lesions, alcohol and drug use did not appear to be associated with polydipsia. A better understanding of polydipsia may help clarify the pathophysiology of schizophrenia and of the most severe cases of this disabling illness.
Acknowledgements This study was made possible by the collaboration of the staff of the hospital and the support of Albert R. Di Dario, Superintendent of Norristown State Hospital at that time and of Richard Jossiasen PhD Director of
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
the Clinical Research Center. Jose de Leon’s salary was partly supported by the NIH Grant MH-51380. Margaret T. Susce, R.N., M.L.T. helped with editing of this article. We would like to thank the third reviewer, whose comments led us to some new analyses, which demonstrated that using different definitions of polydipsia did not influence the results.
References Adler, L.E., Olincy, A., Waldo, M., Harris, J.G., Griffith, J., Stevens, K., Flach, K., Nagamoto, H., Bickford, P., Leonard, S., Freedman, R., 1998. Schizophrenia, sensory gating, and nicotine receptors. Schizophr. Bull. 24, 189 – 202. Blum, A., Tempey, F.W., Lynch, W.J., 1983. Somatic findings in patients with psychogenic polydipsia. J. Clin. Psychiatry 44, 55 – 56. Buckley, P.F., 1998. Substance abuse in schizophrenia: a review. J. Clin. Psychiatry 59 (Suppl. 3), 26 – 30. de Leon, J., Verghese, C., Tracy, J., Josiassen, R., Simpson, G.M., 1994a. Polydipsia and water intoxication in psychiatric patients: a review of the epidemiological literature. Biol. Psychiatry 35, 408 – 419. de Leon, J., Canuso, C., White, A.O., Simpson, G.M., 1994b. A pilot effort to determine benztropine equivalents of anticholinergic medications. Hosp. Comm. Psychiatry 45, 606 – 607. de Leon, J., Dadvand, M., Canuso, C., Odom-White, A., Stanilla, J.K., Simpson, G.M., 1996. Polydipsia and water intoxication in a long-term psychiatric hospital. Biol. Psychiatry 40, 28 – 34. de Leon, J., Tracy, J., McCann, E., McGrory, A., Diaz, F., in press. Schizophrenia and tobacco smoking: a replication study in another US psychiatric hospital. Schizophr. Res. Evenson, R.C., Jos, C.J., Mallya, A.R., 1987. Prevalence of polydipsia among public psychiatric patients. Psychol. Rep. 60, 803 – 807. Goldman, M.B., 1991. A rational approach to disorders of water balance in psychiatric patients. Hosp. Comm. Psychiatry 42, 488 – 494.
301
Goldman, M.B., Robertson, G.L., Luchins, D.J., 1997. Psychotic exacerbations and enhanced vasopressin secretion in schizophrenic patients with hyponatremia and polydipsia. Arch. Gen. Psychiatry 54, 443 – 449. Luchins, D.J., Goldman, M.B., Lieb, M., Hanrahan, P., 1992. Repetitive behaviors in chronically institutionalized schizophrenic patients. Schizophr. Res. 8, 119 – 123. Luchins, D.J., Nettles, K.W., Goldman, M.B., 1997. Anterior media temporal lobe volumes in polydipsic schizophrenic patients with and without hypo-osmolemia: a pilot study. Biol. Psychiatry 42, 767 – 770. Mercier-Guidez, E., Loaz, G., 2000. Polydipsia and water intoxication in 353 psychiatric inpatients: an epidemiological and psychopathological study. Eur. Psychiatry 15, 306 – 311. Ripley, T.L., Millson, R.C., Koczapski, A.B., 1989. Self-induced water intoxication and alcohol abuse. Am. J. Psychiatry 146, 102 – 103. Shah, P.J., Greenberg, W.M., 1992. Polydipsia with hyponatremia in a state hospital population. Hosp. Comm. Psychiatry 43, 509 – 511. Shutty Jr., M.S. 1996. Cigarette use, drinking and voiding in schizophrenic patients with polydipsia and hyponatremia. Schizophr. Res. 21, 195 – 197. Shutty, M.S., McCulley, K., Pigott, B., 1995. Association between stereotypic behavior and polydipsia in chronic schizophrenic patients. J. Behav. Ther. Exp. Psychiatry 26, 339 – 343. Sleeper, F.H., 1935. Investigation of polyuria in schizophrenia. Am. J. Psychiatry 91, 1019 – 1031. Tracy, J., de Leon, J., Querishi, G., McCann, E., Josiassen, R., 1996. Repetitive behaviors in schizophrenia: discrete symptoms or single disturbance. Schizophr. Res. 20, 221 – 229. Vieweg, W.V.R., Weiss, N.M., David, J.J., Rowe, W.T., Godleski, L.S., Spradlin, W.W., 1988. Treatment of psychosis, intermittent hyponatremia, and polydipsia (PIP syndrome) using lithium and phenytoin. Biol. Psychiatry 23, 25 – 30. Vieweg, W.V.R., Godleski, L.S., Hundley, P.L., Harrington, D., Yank, G., 1989. Antipsychotic drugs, lithium, carbamazepine, and abnormal diurnal weight gain in psychosis. Neuropsychopharmacology 2, 39 – 43.
Polydipsia and schizophrenia in a psychiatric hospital: a replication study Jose de Leon a,*, Joseph Tracy b, Eileen McCann a, Amy McGrory a a
Mental Health Research Center at Eastern State Hospital, 627 West Fourth St., Lexington, KY 40508, USA b Thomas Jefferson University, Jefferson Medical College, Philadelphia, PA, USA Received 20 February 2001; accepted 4 June 2001
Abstract A prior study in a US state hospital suggested that schizophrenia, smoking and long hospitalization were associated with polydipsia. This study, in another US hospital, attempts to (1) replicate that schizophrenia and smoking are associated with polydipsia, and (2) rule out that this relationship is partly explained by alcohol and drug use. Both studies have similar methodologies. The second sample included 588 inpatients. Models of variables associated with polydipsia were developed using logistic regression. In the second study, after correcting for other factors, the association between polydipsia and schizophrenia showed a borderline significance, while polydipsia and smoking displayed a significant association. Neither organic brain lesions, nor alcohol or drug use, were associated with polydipsia. An analysis combining both samples showed that: (1) schizophrenia, long hospitalization, smoking and heavy smoking were significantly associated with polydipsia, and (2) male gender and Caucasian race (but not smoking) increased the risk of developing water intoxication in polydipsic patients. These two studies in severely mentally ill patients suggest that the association of polydipsia with schizophrenia, smoking and chronicity is consistent and independent from the definition of polydipsia (by staff, a biological method or the combination of both). Psychiatric medications do not appear to explain most cases of polydipsia in these patients. D 2002 Elsevier Science B.V. All rights reserved. Keywords: Schizophrenia; Polydipsia; Water intoxication; Nicotine; Smoking; Logistic regression
1. Introduction Polydipsia, among chronic psychiatric patients, is poorly understood and underdiagnosed. Patients may have simple polydipsia, with accompanying polyuria, polydipsia, with water intoxication, and/or physical
*
Corresponding author. Tel.: +1-859-246-7487; fax: +1-859246-7019. E-mail address: [email protected] (J. de Leon).
complications secondary to ingestion of fluids in large quantities (de Leon et al., 1994a). Ingestion of lithium or diuretics, and endocrine disorders such as diabetes mellitus, are causes of polyuria, frequently found among chronic psychiatric inpatients. In all of these polyuric conditions, the primary abnormal mechanism is polyuria with a compensatory or ‘‘secondary’’ polydipsia. (For simplification, in this article, those types of etiologies are designated as medical causes of polyuria). On the other hand, the typical polydipsia that induces a
0920-9964/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 0 - 9 9 6 4 ( 0 1 ) 0 0 2 9 2 - 4
294
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
secondary polyuria (seen in psychiatric patients) will be termed ‘‘primary’’ polydipsia. In almost all cases of polydipsia with water intoxication, there is an increased release of antidiuretic hormone (ADH) (Goldman et al., 1997). Alternatively, psychiatric patients without polydipsia may develop a syndrome of inappropriate secretion of ADH (SIADH) induced by drugs such as carbamazepine or sulfonylureas. It is also possible that SIADH may occur in non-polydipsic patients in the absence of unrecognized causes. Hyponatremia in psychiatric patients may also be precipitated by other factors which may decrease free-water excretion, such as thiazide diuretics or hypothyroidism (Goldman, 1991). 1.1. Polydipsia According to a review of the literature, primary polydipsia may be present in more than 20% of chronic inpatients (de Leon et al., 1994a). In a recent French study, the staff reports of 253 chronic patients were reviewed. The study results showed a 13% rate of polydipsia among the sample and a 10% rate of primary polydipsia (Mercier-Guidez and Loaz, 2000). Little is known about what factors contribute to polydipsia. Chronicity of illness (Evenson et al., 1987; de Leon et al., 1996), and surprisingly, smoking (de Leon et al., 1996) may be associated with polydipsia. The literature has suggested that heavy smoking is only associated with the development of water retention and water intoxication in polydipsic patients but a recent French study (Mercier-Guidez and Loaz, 2000) confirmed our findings that there is an association between polydipsia and smoking. 1.2. Polydipsia with water intoxication Previous surveys suggest that an average of 29% (25 – 86%) of patients with polydipsia have a history of water intoxication (de Leon et al., 1994a). However, in our first survey, which diagnosed a larger number of polydipsic patients than in prior studies, the prevalence of a history of water intoxication was only 14% (de Leon et al., 1996). A more recent survey suggested a prevalence of 31% (Mercier-Guidez and Loaz, 2000). Drugs that decrease free-water excretion (thiazide diuretics, sulfonylureas, and carbamazepine) may
contribute to the development of water intoxication in patients with polydipsia. Additionally, Caucasian race, male gender, chronicity, a diagnosis of schizophrenia and smoking may also be associated with the development of water intoxication in patients with polydipsia (Evenson et al., 1987; Shah and Greenberg, 1992; de Leon et al., 1994a, 1996). These factors may influence each other but it is not clear which of these factors may have a primary influence. Our first epidemiological study, which included all inpatients at a psychiatric hospital in Philadelphia, was important in order to establish that polydipsia is associated with schizophrenia and smoking after correcting for the effect of other factors, such as chronicity. No replications of this study have been published using multivariate analyses to correct for the effect of other factors. Moreover, our study did not correct for the effects of abuse of other substances. Some data in the literature suggests that schizophrenic patients with substance abuse disorders demonstrate high rates of polydipsia and water intoxication (Ripley et al., 1989). It has been found that patients with schizophrenia tend to abuse alcohol and drugs more frequently than the normal population (Buckley, 1998). This current study, at a second US psychiatric state hospital near Philadelphia, tried to (1) replicate the data showing that schizophrenia and smoking are associated with polydipsia; and (2) rule out that this relationship is partly explained by abuse of alcohol and/or drugs.
2. Methods The methodology applied to both studies was very similar (de Leon et al., 1996). The first study was performed in 1990 and included all patients admitted at Haverford State Hospital located in suburban Philadelphia. This second replication study was performed in 1992 at Norristown State Hospital, also in suburban Philadelphia. Both hospitals serviced Philadelphia and the surrounding counties but their catchment areas did not overlap. 2.1. Subjects Out of the 659 patients in the hospital, 588 agreed to participate in this study. Of those 588 patients, a few either did not provide all of the requested infor-
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
mation, or cooperated with only some of the procedures. As expected, this affected the sample size in some of our analyses. The sample in this study has been previously described (de Leon et al., in press). Sixty-three percent (370/588) of the patients were male. Seventy-one percent (415/588) of the patients were Caucasian, 27% (161/588) were African-American and 2% (12/ 588) were from other races. The mean patient age was 46 years (standard deviation [S.D.] 16.1). Twentyeight percent (166/588) of the patients were older than 55 years of age. The mean duration of current hospitalization was 9.2 years (S.D. 15.0). Most patients (90%) were taking antipsychotics. Mean dose in chlorpromazine equivalents was 931 (S.D. 845). The percentage of patients with history of alcohol abuse or dependence was 39%. The percentage of patients with history of abuse or dependence on any drug was 31%. Seventy percent of the patients were smokers, and of those, 17% were heavy smokers ( z 1.5 packs/day). Clinical diagnoses were taken from the charts. The DSM-IIIR diagnoses were as follows: schizophrenia (71%), schizoaffective disorder (7%), organic mental disorders (8%), affective disorders (5%) and other disorders (9%). As in the first study, the diagnoses were divided into schizophrenic (schizophrenic or schizoaffective) and non-schizophrenic categories, 78% and 22%, respectively. Antipsychotic medications were converted into chlorpromazine equivalents, with dosages higher than 1000 mg of chlorpromazine equivalents classified as ‘‘high’’. A table was used to convert antimuscarinic activity of psychotropic medication to benztropine equivalents (de Leon et al., 1994b). Dosages equal to or greater than 2 mg of benztropine equivalents were classified as ‘‘moderate’’. 2.2. Assessment of polydipsia The AM weight was obtained between 7 and 8 AM, after the patient voided and before breakfast. After ensuring that patients were in the same state of dress, the PM weight was obtained between 3 and 4 PM. Normalized diurnal weight gain (NDWG), which is the percentage of weight gain from morning to afternoon, was derived by dividing the afternoon weight gain by the morning weight and multiplying by 100 (Vieweg et al., 1989).
295
The NDWG is considered a measure of polydipsia and water retention. In normal individuals, NDWG is lower than 1.2%. An NDWG > 4% is associated with a decrease in serum sodium by at least 10 mM/l and carries a serious risk of water intoxication (Vieweg et al., 1989). AM and PM urine samples were obtained, and the SPGU was measured with a refractometer. SPGU lower than 1.009 is suggestive of polydipsia (Blum et al., 1983). Staff (nurses and nurses’ aides) were also interviewed regarding patients who were drinking excessive amounts of fluid (Luchins et al., 1992; Tracy et al., 1996). 2.3. Definitions of polydipsia and water intoxication A diagnosis of polydipsia by staff or an SPGU lower than 1.009 was considered a positive case of polydipsia (de Leon et al., 1996). Some polydipsic patients may have secondary polydipsia explained by polyuria induced by lithium, diuretics or diabetes mellitus. The presence of any of these causes of polyuria was considered as a confounding variable in the logistic regression models. In the prior study, a history of water intoxication was established by an extensive chart reviewed by a physician, however, this was not an option in this study. To establish which factors were associated with a high risk of hyponatremia among patients with primary polydipsia, those patients were subdivided using the NDWG. Polydipsic patients with very high NDWG (>4.0%) were classified as displaying a very high risk of developing water intoxication. 2.4. Statistics The Statistical Package for Social Sciences (SPSS) was used for calculations. Two-way cross-tabulations for univariate analyses (Table 2) computed the odds ratios (ORs). Univariate analyses are described to compare with other scientific reports that may not use logistic regression. The 95% confidence intervals (CI) for ORs were calculated. Logistic regression analyses of the probability that a patient was polydipsic were corrected by using several independent variables: diagnosis (schizophrenic or non-schizophrenic group), smoking, heavy smoking ( z 1.5 packs/day or not), high dose of antipsychotic medi-
296
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
hospitalization (4.8 years), and dose in chlorpromazine equivalent (732) were also lower. The samples of both hospitals were combined in a logistic regression analysis and the effect of these clinical variables was controlled. The differences between the two hospitals should not affect comparisons between the first and replication studies; the logistic model allows for studying the effect of a particular variable while correcting for the effects of other variables. Logistic regressions that included history of alcohol or drug abuse disorders and CNS abnormalities were performed only in this replication study.
cation (>1000 chlorpromazine equivalents or not), moderate doses of antimuscarinic drugs ( z 2 benztropine equivalents or not), old age (>55 years or not), gender, race, long duration of hospitalization (>2 years or not), and causes of polyuria. Any patient with history of substance abuse or dependence in their chart was considered positive for alcohol or drug disorders. By chart review, history of any ‘‘organic’’ central nervous system (CNS) abnormality was also established in 36% of patients. The Hosmer –Lemeshow Goodness-of-Fit test was used to test the fitness of the logistic models; all fit well. Logistic regression analyses for polydipsia are reported in Table 3. Similar analyses were performed within the polydipsic patient sample. After excluding medical causes of water retention (hypothyroidism or treatment with carbamazepine, sulphonylureas or thiazides), polydipsic patients were divided into those with a high risk of developing water intoxication (NDWG>4.0%) and those without. There were some differences between the first study and this study: the percentage of Caucasians in the first hospital was higher (84%), but the percentage of elderly patients (15%) and patients with schizophrenia (65%) were lower. Additionally, in the first hospital the mean age (42 years), duration of
3. Results 3.1. Polydipsia 3.1.1. Frequency of polydipsia The overall frequency of abnormalities in SPGU was a little higher in this second study in both the total and the schizophrenic sample (Table 1). Thus, the percentage of people with polydipsia, when determined by combining staff report and SPGU, was slightly higher in the second hospital (49%, CI 45–
Table 1 Frequencies of polydipsia and high risk of water intoxication Polydipsia All sample
Staff diagnosis AM SPGU < 1.009 PM SPGU < 1.009 SPGU < 1.009a Combining staff and SPGU No medical causes Medical causes
Schizophrenia
First
Replication study
First
Replication study
26% 20% 29% 34% 42% 19% 23%
24% 33% 43% 47% 49% 34% 15%
33% 33% 35% 39% 48% 23% 25%
28% 36% 46% 50% 53% 37% 16%
(142/588) (139/428) (150/346) (221/466) (290/588) (200/588) (90/588)
(124/449) (117/330) (123/269) (179/359) (237/449) (166/449) (171/449)
Polydipsia with high risk of water intoxication All sample
Very high NDWG (>4.0%) With causes of water retention Without causes of water retention a
Lower of AM and PM SPGUs.
Schizophrenia
First
Replication study
First
Replication study
5.6% 0.3% 5.3%
4.4% (26/588) 0.7% ( 4/588) 3.7% (22/588)
8.0% 0.4% 7.6%
4.7% (21/449) 0.7% (3/449) 4.0% (18/449)
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
53% versus 42%, CI 37– 47%). However, this difference was not significant. 3.1.2. Univariate analyses of polydipsia The results from the univariate analyses were remarkably similar to those of the first study in which schizophrenia, smoking, heavy smoking, and causes of polyuria were also significantly associated with polydipsia. Other variables were not so consistent in both studies (Table 2). 3.1.3. Logistic regression for polydipsia Remarkably, there were many similarities in the variables predicting polydipsia in both studies’ total samples. For instance, causes of polyuria were significant in both studies, as expected; schizophrenia had borderline significance after correcting for other factors; and heavy smoking was significant in the first study but only borderline in the second. Finally, when both studies were combined and the effects of the hospital and the causes of polyuria were corrected, schizophrenia, long hospitalization, smoking and heavy smoking were all significant factors in predicting diagnosis of polydipsia (Table 3). In this second study, we found that smoking was associated with polydipsia in the schizophrenic sample. Additionally, long hospitalization, smoking and heavy smoking were all significant factors after correcting for other variables and combining both hospi-
297
tal samples. In the non-schizophrenic sample, long hospitalization and causes of polyuria were significant in both hospitals. After combining both hospitals, heavy smoking also became significant. 3.1.4. Correcting for other factors In the second study, history of alcohol and drug abuse disorders was never a significant factor in univariate analyses or logistic regression with polydipsia as the dependent variable. The presence of any kind of ‘‘organic’’ CNS abnormality was also non-significant. 3.2. Polydipsia with high risk of developing water intoxication 3.2.1. Frequency of polydipsia with high risk of developing water intoxication There was a slightly lower frequency of polydipsia with water retention in the second study. Note there is no way of estimating how uncooperativeness may have influenced these differences (Table 1). 3.2.2. Univariate analyses of polydipsia with high risk of developing water intoxication In the first study, male gender, a diagnosis of schizophrenia, high doses of antimuscarinic drugs, race and prolonged hospitalizations were associated with high NDWG in patients with polydipsia, after excluding causes of water retention. In this replication
Table 2 Odds ratios (and confidence intervals) of significant variables associated with polydipsia and with high risk of water intoxication in univariate analyses High risk of water intoxication in polydipsic patients
Polydipsia
Schizophrenia Smoking Heavy smoking High antipsychotic dose Antimuscarinic drugs: moderate dose Male gender Old age Caucasian race Long hospitalization Causes of polyuria
First study
Replication study
2.2 2.4 4.3 1.8 1.6
1.8 (1.2 – 2.6) 2.9 (2.0 – 4.2) 2.6 (1.5 – 4.6)
0.54 b.s.
1.5 (1.1 – 2.1) 0.72 (0.50 – 1.0) b.s.
1.9 2.9
1.5 (1.0 – 2.1)
First study
Replication study
4.0 2.8 b.s.
* 4.0 (1.1 – 14.0)
4.3
*
Only significant results are reported in the table. Blank spaces indicate non-significant odds ratios. b.s.: borderline significance. * Not significant but results were in the opposite direction with p-values 0.05 – 0.10.
298
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
Table 3 Odd ratios (and confidence intervals) of significant variables associated with polydipsia in logistic regression models Total sample First study
Schizophrenia Smoking Heavy smoking Old age Long hospitalization Causes of polyuria
Replication study
Combination of both studiesa
N = 360
N = 536
N = 896
1.4 b.s.
1.5 (0.97 – 2.3) b.s. 2.3 (1.5 – 3.4) 1.7 (0.97 – 3.1) b.s.
1.5 (1.1 – 2.1) 2.0 (1.4 – 2.8) 2.5 (1.7 – 3.8)
1.6 (1.1 – 2.3)
1.5 (1.1 – 2.1) 1.9 (1.4 – 2.6)
Replication study
Combination of both studiesa
3.4 0.29 1.9 2.6 Schizophrenic sample First study N = 237
Smoking Heavy smoking Old age Antimuscarinic drugs: Moderate dose Long hospitalization Causes of polyuria
N = 416
N = 652
2.5 (1.5 – 4.0)
2.1 (1.4 – 3.2) 2.3 (1.5 – 3.5)
3.3 0.42 b.s. 1.9 2.1 2.1
1.5 (1.1 – 2.1) 1.5 (1.1 – 2.2)
Non-schizophrenic sample
Smoking Heavy smoking Long hospitalization Causes of polyuria
First study
Replication study
Combination of both studiesa
N = 123
N = 120
N = 243
4.3 2.2 b.s. 5.4
2.2 (0.90 – 5.4) b.s. 3.4 (1.4 – 8.4)
3.8 (1.5 – 9.4) 1.9 (1.0 – 3.5) 4.0 (2.1 – 7.5)
Only significant results are reported in the table. Blank spaces indicate non-significant odds ratios. b.s.: borderline significance. a The second hospital had association with polydipsia: corrected OR was 1.9 (1.4 – 2.6) in the total sample, 1.8 (1.2 – 2.5) in the schizophrenic sample and 2.3 (1.2 – 4.2) in the non-schizophrenic sample.
study, only male gender was associated with high NDWG. Long hospitalizations and moderate doses of antimuscarinic drugs had a borderline significance. However, these two latter factors were associated with low NDWG (Table 2). 3.2.3. Logistic regression for polydipsia with high risk of developing water intoxication Logistic regressions of the first and second hospital showed very similar results to univariate analyses (Table 2). When both samples were combined (allowing greater statistical power) and all variables included in Table 2 were tested, only male gender was significant (OR = 2.6, CI 1.1 –6.3). Caucasian race was close to significance (OR = 3.0, CI 0.97 – 9.3).
4. Discussion 4.1. Limitations This study has several limitations, including a small amount of missing data, the lack of repeated biological measures, and the lack of a thorough chart review of all available records to provide figures of clearly identified episodes of water intoxication (5% in the first study). Missing data in some variables may influence estimations of prevalence, but it is important to remember that this study included a large sample. The staff at both hospitals had known these patients in these studies for many years. We included staff reports in the diagnosis of polydipsia because we believe these
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
practice results in a substantial decrease in false negatives from patients who were not drinking too much the day of the survey. Of course, it is possible that this may led to some false positives, however, based on our observations in both hospitals, the staff tended to report false negatives and they did not identify some of the obvious polydipsic patients who may have been drinking when off the units. However, staff diagnoses appear to have a lower number of false positives of polydipsia in general (staff members do not identify non-polydipsic patients as polydipsics). On the other hand, the staff may not be able to distinguish primary and secondary polydipsia. To ensure that the definition of polydipsia was not an issue, we repeated the analyses after combining both studies using either only staff diagnosis or biological diagnosis. Using only the staff diagnosis, the OR was 1.7 for schizophrenia, 1.8 for smoking, 2.7 for heavy smoking and 2.2 for long hospitalization. Using only the SPGU diagnosis, the OR was 1.5 for schizophrenia, 1.6 for smoking, 2.4 for heavy smoking and 1.4 for long hospitalization. These results are similar to those described in Table 3 using the combined definition of polydipsia (staff and SPGU). It is even more remarkable that two new analyses, using two complete independent and unrelated methods to diagnose polydipsia, staff diagnose and SPGU, came up with the same common variables (schizophrenia, smoking, heavy smoking and long hospitalization) associated with polydipsia. In summary, these two surveys in two US hospitals should be considered unique and valuable data because it is increasingly difficult to amass such surveys since fewer and fewer psychiatric patients spend many years within the confines of a psychiatric hospital. Also, the current ethical research regulations make these noninvasive naturalistic studies more difficult. 4.2. Strengths However, in spite of its limitations, one must remember that our first study is the only epidemiological study of a large and complete sample of severely mentally ill patients, using a multivariate method (logistic regression) to explore which variables are associated with polydipsia. Replication studies are frequently used in scientific methodology to test for the robustness of the scientific findings. Results that tend to be replicated in several
299
studies probably reflect ‘‘true’’ associations. Results that are not replicated in later studies or with different methods may reflect spurious or chance findings. In summary, we believe that the replication, by use of the similar methodology in both studies, helps to strengthen the reliability of the results. Moreover, the combination of both samples provided greater statistical power. Schizophrenia, smoking and chronicity appear to be consistently associated with polydipsia across the tables. This was also true when polydipsia was only defined by the staff or only by SPGU. Psychiatric medications do not appear to explain most cases of polydipsia. However, lithium and medical causes of polyuria appear to be more powerful predictors of polydipsia in non-schizophrenic patients. 4.3. Differences between the two hospitals After correcting for the effect of confounding variables, the odds of polydipsia in the second hospital was 1.8 times higher than in the first (Table 3). The effect of larger chronicity of the patients in the second hospital may not explain this difference since the logistic regression model will correct for it. It cannot be ruled out that a better measure of chronicity (such the unavailable duration of unremitting illness or active psychosis) may explain this difference between these two studies. It is possible that the refractometers for measuring the SPGU had different sensitivities, but this was not tested. Regardless, the SPGU is a relatively ‘‘gross’’ method. Better methods, such as urine osmolality concentration of creatinine, were not available. 4.4. Polydipsia and schizophrenia It is not known why schizophrenia may be associated with polydipsia. The temporal lobe abnormalities that are particularly found in schizophrenic patients have been posited to be important in patients with polydipsia and hyponatremia (Luchins et al., 1997). However, this will not explain simple polydipsia. 4.5. Polydipsia and smoking The association between polydipsia, smoking, and heavy smoking is remarkable. Particularly, it is striking that both have additive factors. Since nicotine increases ADH secretion, it is expected that smoking
300
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
may contribute to water retention and hyponatremia in polydipsic patients (de Leon et al., 1994a). However, these two studies suggest the association appears to be between smoking and increased fluid intake. The physiological literature does not provide clues as to why smoking may be associated with an increased fluid intake. However, Adler et al. (1998) described that the subfornical organ involved in the regulation of water drinking is rich in nicotine receptors. Literature coming from experimental psychology field has led to suggest that both smoking and polydipsia may be repetitive behaviors associated to situations of low reinforcement. The association between polydipsia and other repetitive behavior has been tested in two types of designs: within and across individuals. Using a within-individual design, Shutty et al. (1995) found that in two polydipsic patients there was a temporal relationship with stereotyped grooming and pacing. Using an acrossindividual design in the schizophrenic patients of our second hospital, we found that other repetitive behavior was not related to polydipsia defined by staff but smoking was indeed associated to polydipsia (Tracy et al., 1996). In a carefully controlled study of 10 polydipsic patients with history of hyponatremia, Shutty (1996) found that cigarette use was correlated with the total amount drunk. So in summary, this behavioral literature suggest that smoking and water drinking may be repetitive behaviors that are in some way associated but it is not clear whether polydipsia is associated with other repetitive behaviors or not. 4.6. Polydipsia and psychiatric medications Before any psychiatric medications were available, Sleeper (1935) demonstrated by catheterization procedures that 25% of 92 male schizophrenics were polydipsic. This classical study proves unequivocally that primary polydipsia in schizophrenia cannot be explained by antipsychotic medications. Certainly, polydipsia explained by lithium (or diuretics) may explain some cases of polydipsia, particularly, among non-schizophrenic patients. The presence of lithium treatment does not completely rule out primary polydipsia; moreover, lithium has been used to decrease the risk of hyponatremia in primary polydipsic patients (Vieweg et al., 1988).
4.7. Polydipsia and alcohol and drug abuse Alcohol and drug use did not appear to be associated with polydipsia in the second hospital. Replication from an independent study may be needed. 4.8. High risk of developing water intoxication in polydipsic patients Obviously, polydipsic patients taking any medication that is either associated with SIADH (carbamazepine or sulphonylureas) or decreases free-water clearance (thiazides) is at increased risk of developing water intoxication. In our first study, history of water intoxication by chart review appeared to be associated with longer hospitalization (de Leon et al., 1994a). The prior literature suggested that male gender and Caucasian race (versus African-American race) might also be associated with water intoxication. In our first study, both appear to be overrepresented but did not reach significance in the small sample studied. Both male gender and Caucasian race appeared to be associated with high risk of developing water intoxication when both sets of samples were combined.
5. Conclusion In summary, these two studies in severely mentally ill patients suggest that the association of polydipsia with schizophrenia, smoking and chronicity is consistent and independent from the definition of polydipsia (by staff, a biological method or the combination of both). Psychiatric medications do not appear to explain most cases of polydipsia in these patients. Organic brain lesions, alcohol and drug use did not appear to be associated with polydipsia. A better understanding of polydipsia may help clarify the pathophysiology of schizophrenia and of the most severe cases of this disabling illness.
Acknowledgements This study was made possible by the collaboration of the staff of the hospital and the support of Albert R. Di Dario, Superintendent of Norristown State Hospital at that time and of Richard Jossiasen PhD Director of
J. de Leon et al. / Schizophrenia Research 57 (2002) 293–301
the Clinical Research Center. Jose de Leon’s salary was partly supported by the NIH Grant MH-51380. Margaret T. Susce, R.N., M.L.T. helped with editing of this article. We would like to thank the third reviewer, whose comments led us to some new analyses, which demonstrated that using different definitions of polydipsia did not influence the results.
References Adler, L.E., Olincy, A., Waldo, M., Harris, J.G., Griffith, J., Stevens, K., Flach, K., Nagamoto, H., Bickford, P., Leonard, S., Freedman, R., 1998. Schizophrenia, sensory gating, and nicotine receptors. Schizophr. Bull. 24, 189 – 202. Blum, A., Tempey, F.W., Lynch, W.J., 1983. Somatic findings in patients with psychogenic polydipsia. J. Clin. Psychiatry 44, 55 – 56. Buckley, P.F., 1998. Substance abuse in schizophrenia: a review. J. Clin. Psychiatry 59 (Suppl. 3), 26 – 30. de Leon, J., Verghese, C., Tracy, J., Josiassen, R., Simpson, G.M., 1994a. Polydipsia and water intoxication in psychiatric patients: a review of the epidemiological literature. Biol. Psychiatry 35, 408 – 419. de Leon, J., Canuso, C., White, A.O., Simpson, G.M., 1994b. A pilot effort to determine benztropine equivalents of anticholinergic medications. Hosp. Comm. Psychiatry 45, 606 – 607. de Leon, J., Dadvand, M., Canuso, C., Odom-White, A., Stanilla, J.K., Simpson, G.M., 1996. Polydipsia and water intoxication in a long-term psychiatric hospital. Biol. Psychiatry 40, 28 – 34. de Leon, J., Tracy, J., McCann, E., McGrory, A., Diaz, F., in press. Schizophrenia and tobacco smoking: a replication study in another US psychiatric hospital. Schizophr. Res. Evenson, R.C., Jos, C.J., Mallya, A.R., 1987. Prevalence of polydipsia among public psychiatric patients. Psychol. Rep. 60, 803 – 807. Goldman, M.B., 1991. A rational approach to disorders of water balance in psychiatric patients. Hosp. Comm. Psychiatry 42, 488 – 494.
301
Goldman, M.B., Robertson, G.L., Luchins, D.J., 1997. Psychotic exacerbations and enhanced vasopressin secretion in schizophrenic patients with hyponatremia and polydipsia. Arch. Gen. Psychiatry 54, 443 – 449. Luchins, D.J., Goldman, M.B., Lieb, M., Hanrahan, P., 1992. Repetitive behaviors in chronically institutionalized schizophrenic patients. Schizophr. Res. 8, 119 – 123. Luchins, D.J., Nettles, K.W., Goldman, M.B., 1997. Anterior media temporal lobe volumes in polydipsic schizophrenic patients with and without hypo-osmolemia: a pilot study. Biol. Psychiatry 42, 767 – 770. Mercier-Guidez, E., Loaz, G., 2000. Polydipsia and water intoxication in 353 psychiatric inpatients: an epidemiological and psychopathological study. Eur. Psychiatry 15, 306 – 311. Ripley, T.L., Millson, R.C., Koczapski, A.B., 1989. Self-induced water intoxication and alcohol abuse. Am. J. Psychiatry 146, 102 – 103. Shah, P.J., Greenberg, W.M., 1992. Polydipsia with hyponatremia in a state hospital population. Hosp. Comm. Psychiatry 43, 509 – 511. Shutty Jr., M.S. 1996. Cigarette use, drinking and voiding in schizophrenic patients with polydipsia and hyponatremia. Schizophr. Res. 21, 195 – 197. Shutty, M.S., McCulley, K., Pigott, B., 1995. Association between stereotypic behavior and polydipsia in chronic schizophrenic patients. J. Behav. Ther. Exp. Psychiatry 26, 339 – 343. Sleeper, F.H., 1935. Investigation of polyuria in schizophrenia. Am. J. Psychiatry 91, 1019 – 1031. Tracy, J., de Leon, J., Querishi, G., McCann, E., Josiassen, R., 1996. Repetitive behaviors in schizophrenia: discrete symptoms or single disturbance. Schizophr. Res. 20, 221 – 229. Vieweg, W.V.R., Weiss, N.M., David, J.J., Rowe, W.T., Godleski, L.S., Spradlin, W.W., 1988. Treatment of psychosis, intermittent hyponatremia, and polydipsia (PIP syndrome) using lithium and phenytoin. Biol. Psychiatry 23, 25 – 30. Vieweg, W.V.R., Godleski, L.S., Hundley, P.L., Harrington, D., Yank, G., 1989. Antipsychotic drugs, lithium, carbamazepine, and abnormal diurnal weight gain in psychosis. Neuropsychopharmacology 2, 39 – 43.