LARSON, Justice.

Five workers in a Department of Human Services (DHS) office building in Ottumwa became ill, allegedly the result of a contaminated atmosphere in the building. A combination of factors, including the use of an insecticide, sewer gas, and poor ventilation were to blame, according to their claims. Following a ten-week trial, the jury returned substantial verdicts for the plaintiffs. On the defendants' posttrial motions to dismiss and for judgment notwithstanding the verdict, the court vacated its earlier judgments for the plaintiffs. The plaintiffs appealed, and two of the defendants cross-appealed. We affirm in part and reverse in part on the appeal and affirm on the cross-appeals.

Plaintiffs Garnet Bloomquist, Phyllis Mondanaro, Joy Miller, and Patsy Watson were employees of the State of Iowa working for DHS in its Ottumwa office. Linda Owens, an employee of Wapello County, worked in the same building. For purposes of ruling on the motion for judgment notwithstanding the verdict, we view the evidence in the light most favorable to the plaintiffs.

Prior to 1980, DHS and the county employees working with them were located in the Wapello County Courthouse. When these offices became crowded, the county purchased a 100-year-old building and remodeled it. The workers moved in in 1980. A serious infestation of fleas moved in at the same time.

Soon after the employees moved in, they began to complain about the quality of the air and the fleas. A spraying program was commenced by defendant Atomic Termite Pest Control. The spray, an organophosphate called Dursban, was applied in a "crack and crevice" method, on a monthly basis. However, in 1984, more fleas moved in. Atomic Termite began to broadcast spray in the building and continued to apply the spray to the cracks and crevices. According to the plaintiffs' evidence, the spray was broadcast over the carpet, in violation of established standards of care, while workers were still in the building. In addition, the offices were sprayed without removing papers being used by the employees, and the workers who were gone during the spraying were allowed to return to the building while the carpet was still wet.

In 1986, an inspection revealed that there were additional problems in the building. The inspectors found an open sewer line in the basement, a leak in a sewer line, floor drains with no traps, floor drains with no clean-out plugs, floor drains that open to the sewer line, a two-inch opening in the stack open to the sewer line, and a lack of combustion air to the boiler in the hot water heater.

It was eventually discovered that the fresh air intake on one of the air handling units on the roof had never been connected, seriously reducing the amount of fresh air taken into the building. In addition, while the rooftop ventilating units were designed to provide intake of fresh air, controls on them were set to reduce the intake of fresh air below acceptable standards; the reason was to reduce the cost of heating and cooling outside air.

On repeated occasions, workers in the building complained to the state and to the county, asking them to inspect the building and determine the cause of the workers' problems. According to the plaintiffs, these complaints went largely unheeded. Too costly, according to the responses.

Tests done on the carpet in the building several months after the carpet was sprayed revealed that there were still residues of pesticide. It was recommended by Dr. Nyle Kauffman, an internist from Iowa City, that the carpet be removed. The county refused, but it did have the carpet shampooed. The residue from the pesticide remained.

The plaintiffs sought medical assistance. All of the workers were initially treated by Dr. Kauffman, who observed several similarities of the plaintiffs' symptoms. Their problems basically consisted of respiratory problems, immune problems, brain damage, urinary incontinence, and fecal incontinence.

Other doctors, including Dr. Frank Gersh, a neuropsychologist, and Dr. Vernon Varner, a psychiatrist, concurred in Dr. Kauffman's diagnoses. Two of the plaintiffs, Miller and Owens, were treated by Dr. Marc Hines, a certified neurologist in Ottumwa. Dr. Hines diagnosed them as suffering from chronic organophosphate poisoning.

The plaintiffs' experts concurred in their opinions that medical problems experienced by the plaintiffs were permanent and were caused by the conditions present in the DHS building, primarily the flea spray.

In sustaining the defendants' motions for judgment notwithstanding the verdict, the district court ruled as a matter of law that the plaintiffs had failed to establish proximate cause, failed to establish a duty running from the state to the respective plaintiffs, and failed to establish subject matter jurisdiction in the loss-of-consortium claims by plaintiff Owens' children.

I. The Proximate Cause Issue.

The most difficult issue in this case, and one that is common to all of the plaintiffs' claims, is whether the plaintiffs sufficiently established a proximate cause between the conditions in the DHS building and the resulting injuries. Most of the plaintiffs' evidence of causation, produced through their experts' testimony, was admitted without objection. The defendants argue, nevertheless, that this case, which is commonly referred to as a "toxic tort" case, requires more than conventional evidence of causation. The health effects of Dursban, the chief problem in the DHS office, are not fully understood, and the plaintiffs' evidence of proximate cause was merely speculative. In such a case, "epidemiological" evidence is required, according to the defendants. The district court agreed, observing that the plaintiffs' evidence of causation was nothing "other than unproven medical speculation, which is not accepted by mainstream medicine." The court noted that, in any event, any probative value in the plaintiffs' evidence was outweighed by its prejudicial effect. See Iowa R.Evid. 403.

The key issue is whether we will recognize as sufficient, in toxic tort cases, proof of causation based on traditional cause-and-effect testimony, such as by treating doctors, or whether we will require epidemiological evidence as suggested by the defendants and required by the district court.

One of the best known fields of toxic tort litigation involves a drug called Bendectin, an antinausea drug taken by pregnant women. Bendectin has been alleged in many suits to have caused severe birth defects in children. The federal circuits have split on the question of whether causation may be established without epidemiological evidence. See Turpin v. Merrell Dow Pharmaceuticals, Inc., 959 F.2d 1349, 1351-53 (6th Cir.), cert. denied, ___ U.S. ___, 113 S.Ct. 84, 121 L.Ed.2d 47 (1992) (discusses divergent views, among federal circuits, on toxic tort cases and epidemiological evidence).

The district court, in finding no proximate cause as a matter of law, relied on Brock v. Merrell Dow Pharmaceuticals, Inc., 884 F.2d 167 (5th Cir. 1989) (second opinion, en banc). Brock was a Bendectin case, and the court ruled that, in the present state of knowledge about Bendectin, the conventional proof of causation presented by the plaintiffs was insufficient. The Brock court, in its earlier opinion, had suggested that, as more is known about Bendectin, the rule might be relaxed:

Assuredly, one day in the future, medical science may have a clearer understanding of the mechanics of tissue development in the fetus. However, that is not the case today, and speculation unconfirmed by epidemiologic proof cannot form the basis for causation in a court of law.

Brock v. Merrell Dow Pharmaceuticals, Inc., 874 F.2d 307, 315 (5th Cir. 1989) (panel opinion).

Epidemiology is an accepted scientific discipline dealing with the integrated use of statistics and biological/medical science to identify and establish the causes of human disease. Through epidemiology it is possible to assess a percentage of the risk of a disease that is properly attributable to a given factor such as exposure to an allegedly harmful substance. Through the use of epidemiologic standards, courts are provided with a rational and consistent means for evaluating evidence of a causal relationship between exposure to a particular factor and the incidence of a disease. B. Black D. Lilienfeld, Epidemiologic Proof in Toxic Tort Litigation, 52 Fordham L.Rev. 732, 736 (1984).

An epidemiological study attempts to determine the incidence of certain problems with the exposure of the patient to certain alleged causes. Epidemiological studies do not provide direct evidence that a particular victim was injured by exposure to a substance. They have the potential, however, of generating circumstantial evidence of cause and effect. DeLuca v. Merrell Dow Pharmaceuticals, Inc., 911 F.2d 941, 945 (3d Cir. 1990). Epidemiological studies involve

a process which "amounts to an attempt to falsify the null hypothesis and by exclusion accept the alternative." The null hypothesis is the hypothesis that there is no association between two studied variables; in this case the key null hypothesis would be that there is no association between [the suspected] exposure and [the alleged effects]. The important alternative hypothesis in this case is that [the suspected agent's] use is associated with [the plaintiff's problem].

Id. (citations omitted).

According to one authority, the most important generalized evidence of causation of illnesses whose cause is incompletely understood is epidemiology. Michael Dore, A Commentary on the Use of Epidemiological Evidence in Demonstrating Cause-in-Fact, 7 Harv.Envtl.L.Rev. 429, 430-31 (1983) [hereinafter Dore]. One example of such studies involve the effect of cigarette smoking in causing lung cancer. Id. However, as Dore notes, epidemiological evidence is often inaccurate. For example, if the normal incidence of leukemia of 107 per 100,000 population is doubled, by radiation on the job, to an additional 107 per 100,000 population, the question in a particular case would still be whether the plaintiff contracted leukemia from the radiation on the job or from some other source. Dore, at 436-37. Another problem is that it tends to be too complex for juries to understand. Id. at 437.

This potential for confusing a jury is evident in this discussion of the complex study of epidemiology:

Significance testing has a "P value" focus; the P value "indicates the probability, assuming the null hypothesis is true, that the observed data will depart from the absence of association to the extent that they actually do, or to a greater extent, by actual chance." If P is less than .05 (or 5%) a study's finding of a relationship supportive of the alternative hypothesis is considered statistically significant, if P is greater than. 5% the relationship is rejected as insignificant. Accordingly, the results of a particular study are reported as simply "significant" or "not significant" or as P .05 or P .05.

DeLuca, 911 F.2d at 947 (citation omitted).

There is a risk that the difference between epidemiologist testimony and the testimony of treating doctors might be confused by the jury, Dore, at 437, and experts must make numerous subjective decisions in choosing the control population, evaluating the underlying data, and interpreting the results. The best that can be said for epidemiology is that it can prove the risk but cannot prove individual causation. Dore, at 440.

The DeLuca court discussed this problem in the context of Bendectin exposure. As it noted,

a showing of increased risk for birth defects among women using Bendectin in a particular study does not automatically prove that Bendectin use creates a higher risk of having a child with birth defects because the discrepancy between the exposed and unexposed groups could be the product of chance resulting from the use of only a small sample of the relevant populations.

Id. at 946 (footnote omitted).

Another problem in epidemiological studies is that they require large numbers of patients whose backgrounds are similar in every respect except their exposure to the suspected substance. When studies are made of only a small sample of the relevant population, a so-called "sampling error," the validity of the epidemiological study is brought into question. Id. at 946-47.

To be effective, an epidemiological study requires a large number of cases. For example, in one epidemiological study involving Bendectin, 2218 "pairs" of women were studied. Oxendine v. Merrell Dow Pharmaceuticals, Inc., 506 A.2d 1100, 1107 (D.C.App. 1986). If we were to require epidemiological evidence in all cases of toxic tort injury, we would automatically deny recovery to all claimants who are injured by a toxic substance that is relatively new and as to which a statistical track record has not yet been fully established. We decline to apply a per se requirement of epidemiological evidence simply because of a lack of substantial numbers of cases. The lack of similar cases, of course, would affect the weight of the plaintiffs' evidence of proximate cause.

In our view, while epidemiological evidence is helpful, it should not be held to be an absolute requirement in establishing causation. See Ferebee v. Chevron Chem. Co., 736 F.2d 1529, 1535-36 (D.C.Cir. 1984), cert. denied, 469 U.S. 1062, 105 S.Ct. 545, 83 L.Ed.2d 432 (claim for wrongful death allegedly caused by Paraquat; court held that neither epidemiologic or animal study evidence required if proper methodology is implemented), but see Christophersen v. Allied-Signal Corp., 939 F.2d 1106, 1115 (5th Cir. 1991) (en banc), cert. denied, ___ U.S. ___, 112 S.Ct. 1280, 117 L.Ed.2d 506 (1992) (overruled earlier division case; applied restrictive rule on acceptance of scientific evidence). As the court in Oxendine noted, epidemiological evidence is not necessarily persuasive on either side of an issue; different experts looking at the same studies may come to different conclusions. See Oxendine, 506 A.2d at 1110.

We reject the reasoning of Brock, and therefore the reasoning of the district court in its posttrial ruling, for several reasons: We have long been committed to the principle that issues of proximate cause are only rarely decided as a matter of law. Further, the alleged harmful substances in the Bendectin cases and the present case are substantially different. The amount of knowledge available as to Bendectin, the drug in question in Brock, is relatively limited when compared to the body of knowledge available regarding Dursban. Ancestors of Dursban have long been known and were in fact used to exterminate prisoners in Nazi Germany.

In State v. Hall, 297 N.W.2d 80 (Iowa 1980), this court rejected the "Frye" requirement of "general scientific acceptance" as a prerequisite for introduction of scientific evidence if the reliability could otherwise be established. Id. at 85. See Frye v. United States, 293 F. 1013, 1014 (D.C.Cir. 1923).

Our long-standing policy of liberally admitting expert testimony is now incorporated in our present rules of evidence. Iowa Rule of Evidence 702 provides:

If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training, or education may testify thereto in the form of an opinion or otherwise.

In addition, Iowa Rule of Evidence 703 provides:

The facts or data in the particular case upon which an expert bases an opinion or inference may be those perceived by or made known to him at or before the trial or hearing. If of a type reasonably relied upon by experts in the particular field in forming opinions or inferences upon the subject, the facts or data need not be admissible in evidence.

In light of our liberal rules regarding admission of expert testimony and our view that such evidence would assist the trier of fact, the evidence presented by the plaintiffs in support of their proximate cause claim was properly admitted. The jury found that the evidence was sufficient on causation, and the court erred in deciding otherwise, as a matter of law, simply on the ground that no epidemiological evidence was presented.

The court concluded that, even if the plaintiffs' evidence was probative, any probative value was outweighed by the prejudice of its introduction. See Iowa R.Evid. 403. We conclude for the reasons just discussed, however, that the court's ruling was prompted by an erroneous application of the law, not on the exercise of its discretion.

We hold there was sufficient evidence of causation to generate a jury issue, and the court erred as a matter of law in concluding otherwise simply on the basis that epidemiological evidence was not a part of the record. We therefore reverse on the judgment notwithstanding the verdict as to all of the plaintiffs, except the Owens' children, whose claims are discussed in Division V.

II. The Duty of Care Owed by the State.

Plaintiffs Garnet Bloomquist, Joy A. Miller, Phyllis Mondanaro, and Patsy Watson were employees of the State, and plaintiff Linda Owens was an employee of Wapello County. In their respective suits, because of the exclusive remedy provisions of our workers' compensation law, the state employees were precluded from suing the State and the county employee, Owens, was precluded from suing the County. Therefore, the suits were by the state employees against the County, and the county employee against the State.

The district court sustained the State's motion for judgment notwithstanding the verdict, in part, on the ground that the State owed no duty to Owens; Wapello County was responsible for the building, and the County, rather than the State, had placed Owens in the building. We view the evidence in the light most favorable to Owens.

The director of the Ottumwa Office of the Department of Human Services was a state employee, Frank Jordan. Jordan testified that he had met with the Wapello County Board of Supervisors in order to obtain new quarters for state employees working in DHS. From the time the building was opened, most of the employees were state employees. The State had the discretion as to where the employees were located within the building, and Jordan admitted that he was the person "chiefly responsible" for the health and safety of the workers.

Erma Miller, a state employee in the building, testified that many of the building's workers complained about flea bites, breathing difficulties, dizziness, fatigue, dry eyes, and vomiting. She so informed Jordan, who told her to arrange for Atomic Termite to treat the flea problem. It is the chemical used in this spray that is alleged to be the chief culprit in causing the plaintiffs' illness.

Restatement (Second) of Torts section 343 (1965) provides:

A possessor of land is subject to liability for physical harm caused by his invitees by a condition on the land if, but only if, he

(a) knows or by the exercise of reasonable care would discover the condition, and should realize that it involves an unreasonable risk of harm to such invitees, and

(b) should expect that they will not discover or realize the danger, or will fail to protect themselves against it, and

(c) fails to exercise reasonable care to protect them against the danger.

Restatement (Second) of Torts section 332 (1965), in turn, defines an "invitee" as:

(1) An invitee is either a public invitee or a business invitee.

(2) A public invitee is a person who is invited to enter or remain on land as a member of the public for a purpose for which the land is held open to the public.

(3) A business visitor is a person who is invited to enter or remain on land for a purpose directly or indirectly connected with business dealings with the possessor of the land.

Restatement (Second) of Torts section 328E (1965) defines a "possessor of land" as:

(a) a person who is in occupation of the land with intent to control it or

(b) a person who has been in occupation of land with intent to control it, if no other person has subsequently occupied it with intent to control it, or

(c) a person who is entitled to immediate occupation of the land, if no other person is in possession under Clauses (a) and (b).

Comment d of section 343 states that "[a]n invitee is entitled to expect that the possessor will take reasonable care to ascertain the actual condition of the premises and, having discovered it, either to make it reasonably safe by repair or to give warning of the actual condition and the risk involved therein."

We believe the evidence, when viewed in the light most favorable to the plaintiffs, established a duty of care under section 343 of the Restatement, and hold that it was error for the district court to enter the judgment notwithstanding the verdict on behalf of the State on that ground.

III. The Claim Against Atomic Termite.

As an alternative ground for granting the motion for judgment notwithstanding the verdict on behalf of Atomic Termite, the court stated that it was "virtually uncontradicted" that Atomic Termite had "breached no duty to the plaintiffs in this case since the evidence established that an approved pesticide was applied in an approved manner just as it is done hundreds of times each month in Wapello County."

Again, the trial court was required to view the evidence most favorable to the plaintiffs. When viewed in that light, there was evidence that large amounts of Dursban had been sprayed in the building, even when employees were present. The ventilation system, which was in part inoperative, was not checked. There was evidence that the chemical was applied in such quantities that the carpeting was still damp several days after the spray was applied. According to accepted practices, all treated areas should be vacated during the application of the spray and the spray should never be applied in a broadcast manner with people present. The court erred in setting aside the judgment against Atomic Termite.

IV. The Nursing Expenses.

The trial court also sustained the motions for judgment notwithstanding the verdict insofar as they allowed recovery for "future nursing and related expenses" on the ground that there was insufficient evidence that such services would be required.

The plaintiffs complain that none of the defendants raised this issue in their motions for directed verdict. Moreover, they argue that there was sufficient evidence to support a recovery for future nursing and related expenses. Dr. Kauffman, the internist treating the plaintiffs, testified that the illnesses were significant and long lasting. He testified that Garnet Bloomquist would not be able to care for herself and would require assistance in her daily living. As to plaintiff Linda Owens, he testified that she would require assistance in her daily living.

Other evidence established a poor prognosis for the plaintiffs. We believe it was well within the parameter of the issues and evidence to submit the matter of future nursing services as a means of aiding the plaintiffs in their future care. We reverse on this assignment of error.

V. Loss-of-Consortium Claims by Owens' Children.

The district court, after trial, dismissed the claim by Linda Owens' two children in their loss-of-consortium suit against the State. The jury had returned a verdict in favor of each of the two children.

The district court, in its posttrial ruling, held that it lacked subject matter jurisdiction of the loss-of-consortium claims because the children's claims had not been filed with the State Appeal Board.

Linda Owens' tort claim was filed with the Appeal Board on December 16, 1987; however, her claim did not mention a claim of loss of consortium for her children. When Linda sued the State, she did not mention a loss-of-consortium claim by her children, although her petition did mention that the children had suffered a loss of consortium. No separate damages were requested, and no separate claim was asserted on their behalf.

It is clear that a suit against the State may be maintained only if the claimant has first filed a tort claim with the State Appeal Board. See Iowa Code § 25A.5 (1989) ("No suit shall be permitted under this chapter [State Tort Claims Act] unless the state appeal board has made final disposition of the claim. . . .").

If a claim has not been filed with the appeal board, a court in a subsequent suit lacks subject matter jurisdiction. Brumage v. Woodsmall, 444 N.W.2d 68, 70 (Iowa 1989). Subject matter jurisdiction, of course, cannot be conferred by waiver or consent. Raising the issue in a posttrial motion was therefore proper.

The Federal Tort Claims Act is quite similar to ours, and federal cases under analogous circumstances have held that failure to raise a separate claim through the administrative process deprives the court of jurisdiction. See, e.g., Rucker v. United States Dep't of Labor, 798 F.2d 891, 893 (6th Cir. 1986).

It is argued on behalf of the children that the practical purpose of the notice under the statute is to give the State an opportunity to meet the claim, and it was clear from the State's denial of any liability that it would not have done any good to identify separate claims. We believe that in view of the fact that consortium claims are separate claims under Iowa law it is incumbent on the claimant to first file a claim before the proper administrative agency, even if as a practical matter it would not have changed the State's approach to the case. Under Iowa law, a child's claim for loss of consortium is an independent cause of action. See Audubon-Exira Ready Mix, Inc. v. Illinois Cent. Gulf R.R., 335 N.W.2d 148, 151-52 (Iowa 1983). We agree with the district court that it lacked subject matter jurisdiction of the consortium claims.

VI. The County's Cross-Appeal.

The County cross-appeals from the district court's refusal to grant its motion for a directed verdict on the ground that it owed no duty to the state employees. We conclude that, for the same reasons that the State of Iowa was found to owe a duty under Division II, Wapello County owed a duty of care as the possessor of land, along with the State. It was error to grant the judgment notwithstanding in favor of the County.

VII. Disposition.

On the plaintiffs' appeal, we affirm that part of the district court judgment that dismissed the loss-of-consortium claims on behalf of Linda Owens' children. We reverse as to the balance of the posttrial order and remand for reinstatement of the remaining verdicts other than those originally entered on behalf of the Owens children. In entering judgment, the district court should keep in mind that both the State and the County in their respective cases were found to be more than fifty percent at fault, and therefore principles of joint and several liability should apply. See Iowa Code § 668.4. While this is raised as a separate issue on appeal, we do not address it because the district court will do so on remand.

The plaintiffs' appeal, and Winger Contracting Company's cross-appeal, also challenged the action of the trial court in setting aside a settlement between the plaintiffs and Winger Contracting on the ground that the settlement had not yet been approved by the industrial commissioner. See Iowa Code § 85.22(3). The record before us is inadequate to determine exactly what the trial court did in this respect or to determine whether such approval has since been given. The parties are directed to resolve this issue, following remand, in district court.

Costs are taxed to the appellees.

AFFIRMED IN PART, REVERSED IN PART, AND REMANDED ON APPEAL; AFFIRMED ON CROSS-APPEAL.